Host-transposon mutualism supports regeneration in planarians

Summary

Transposons make up significant fractions of eukaryotic genomes. They are commonly viewed as selfish elements that are detrimental to their hosts, and they are prime targets of specialized host defenses that constrain their expansion. Mutualistic interactions, in which elements co-exist and benefit each other, have so far not been found between transposable elements and eukaryote hosts.

Here, we present evidence for an active transposon that confers a direct benefit to its planarian host. We find that the Ty3-like giant transposon Burro1 is an ancient element that retains its mobility. Burro1 has incorporated a host-derived anti-apoptotic protein that upregulates upon stress and improves stem cell resilience, resulting in enhanced regenerative abilities of its host. Apart from the surprising finding of a transposon’s involvement in planarian regeneration, our data also uncover a true mutualistic interaction between a transposon and a eukaryote.

Graphical Abstract

eTOC blurb

Lee et al. find that the planarian transposon Burro1 bolsters stem cell resilience by producing a functional host-derived caspase-inhibitor during physiological stress. By suppressing apoptosis, active Burro1 elements enhance the regenerative capacity of planarians, and have sustained this mutualistic relationship for over 100 million years.

Introduction

Transposons are widely present in the genomes of eukaryotes and can occupy as much as 85% of the host genome ^1,2. They are commonly viewed as selfish elements that expand and evolve without any particular benefit to their host ^3,4. In fact, they add genomic ballast, their repetitive nature can facilitate undesired recombination events, and their capacity to jump to new genomic locations can disrupt existing elements, supporting their reputation as a threat to their host’s genomic integrity ⁵. Eukaryotes employ a range of mechanisms to defend against transposon activity, and the interaction between transposons and their hosts has been described as an arms race, in which the host fights to inactivate the transposon while the transposon develops an escape or dies ⁶.

This adversarial relation between transposons and eukaryotes forms a stark contrast with the situation in prokaryotes, where mutually beneficial arrangements between mobile genetic elements and their hosts are commonplace. To compensate for their negative effects, prokaryotic plasmids frequently carry genes that improve the host resistance against heavy metals and antibiotics, or improve the acquisition of nutrients ^7-10. Prokaryote composite transposons are known to carry similar genic contributions ^11,12, thereby forming a reciprocal bond with their host where each provides a direct benefit to the fitness of the other that outweighs the burden of the partnership. Such an enduring and reciprocal mutualistic interaction favors not only the survival and propagation of the host, but also the persistence of the functional transposon.

Eukaryotic hosts have adapted to the presence of transposons in their genomes, and in some cases make good use of transposon-derived sequences ¹³, but surprisingly no true reciprocal mutualisms have been found. The chromatin organization of many eukaryotes for example uses the abundance of transposons around telomeres and centromeres to form heterochromatic regions, and uses transposon-contributed CTCF binding sites ^14-20. Additionally, transposons form a significant source of novel genetic material. Their sequences may introduce new enhancer elements that modify expression of neighboring genes ^21-24, and transposon-derived (fusion) proteins have taken on essential functions or have enabled evolutionary novelties including the telomere elongating enzyme TERT, the kinetochore protein CENP-B, the histone methyltransferase Setmar, the RAG genes that mediate VDJ recombination in the vertebrate immune system, and the cell-cell fusion protein Syncytin that is essential for placental development ^25-36. While derived from transposons, these sequences have been domesticated: they no longer mobilize and the remaining (incomplete) sequences are under host control. Such beneficial effects thus may shift the balance towards tolerance of transposons in eukaryote genomes over evolutionary time scales, but they do not contribute to the retention of active transposable elements. Recently, some cases have been reported of active transposons intertwined with the host physiology, but these are thought to reflect addictions: instances where a host process was initially disrupted by the transposon, and the host adaptation to bypass this can only function in the presence of the transposon ^6,37-40. While this makes the transposon indispensable, it provides no net benefit to the host. Furthermore, in each of these cases the repurposed sequences are core transposon genes required for their replication and constitute no fitness cost to the transposon, thereby upholding their reputation as purely selfish elements.

Genome sequencing of the planarian Schmidtea mediterranea recently uncovered a remarkable transposon family called Burro (big unknown repeat rivalling ogre) that consists of three LTR-transposons of giant proportions ⁴¹. Burro transposons belong to the Ty3 superfamily (also known as Gypsy or Metaviridae). Burro elements are between 30 and 35kb long and contain several additional open reading frames (ORFs) besides the Ty3 polyprotein that comprises the essential proteins for mobilization ^42,43. The additional ORFs are unrelated to core transposon genes, suggesting that they have no role in the replication or insertion of the transposon itself, but their function has remained unresolved.

One of the Burro elements, Burro1, is remarkable not only due to its size, but also because it has achieved a high copy number and takes up a significant fraction of the S. mediterranea genome ⁴¹. Here we set out to analyze the life history of this successful giant transposon and its additional ORFs as well as the biological relevance of these additional sequences.

Results

Burro1 is an ancient, highly prolific transposon

The 35 kb consensus sequence of Burro1 consists of one large Open Reading Frame (ORF) encoding the Ty3 polyprotein, which includes the protease, reverse transcriptase (RT), ribonuclease H (RNase H) and integrase, as well as a 5' extended region (Figure 1A) ^41-43. The polyprotein is flanked at the 3' end by several smaller ORFs. The 3' and 5' termini of the transposon consist of 5 kb Long Terminal Repeats (LTRs). The sequence of the core Ty3 polyprotein is similar between Burro1, Burro2, and Burro3, but the additional ORFs 3' of the polyprotein, as well as the 5' extended regions of the polyprotein and the LTRs are distinct between the three family members (Figure S1A).

Figure 1. Life history of Burro1.

(A) Schematic of ORFs (arrows) and conserved protein domains (boxes) in Burro1. Burro1 contains several downstream ORFs, some of which contain conserved protein domains. (B) The Schmidtea mediterranea genome contains both RNA transposons (class I) and DNA transposons (class II). Burro1 (B1) is the most prominent element in the genome. Also highlighted are Burro2 (B2) and Burro3 (B3). (C) qPCR quantification of several Ty3 elements in the S. mediterranea genome. The estimated copy number of Burro1 (left) is in line with the expectation based on the genome assembly (right). n=3 independent samples of 20 animals each. (D) Sequence divergence (Kimura distance) is shown as a measure of age of the most abundant Ty3 elements. Line indicates mean. Burro1 (B1) has copies with a wide range of divergences, indicating that it is one of the oldest transposons in the genome and remained active over a long period. Transposons highlighted in panels F and G are marked in bold. Only full-length elements (90-110% of consensus length) were included in the analysis. (E) Phylogenetic tree of platyhelminths, showing the presence and absence of Burro1, indicating that Burro probably originated within the Geoplanoidea. (F) Scatter plot with linear regression showing the correlation between retained copy number and sequence divergence (age) among Ty3 elements. Burro1 is an outlier that has retained a higher number of copies than would be expected based on its estimated age. (G) Similarity of 5' and 3' LTRs among retained copies of Ty3 elements. Median values are indicated in blue. Existence of Burro1 copies with identical LTRs suggests recent integrations. (H) Schematic of the hypothesized origination of the Burro1 element (see Supplemental Text for further detail).

See also Supplemental Figures S1,S2.

Ty3 retrotransposons comprise over 25% of the Schmidtea genome (Figure 1B). Among these, Burro1 has been particularly prolific: it covers an estimated 8% of the gnomic sequence, making it the single most prominent element. While most Burro1 copies are truncated, based on the current assembly 289 intact copies remain (Figure S1B). Because presence of such long repetitive elements, complicates genome assembly ⁴¹, we used qPCR as an assembly-independent method to estimate the number of Burro1 copies in the Schmidtea genome. Using multiple independent primer sets, we found that the total number of Burro1 copies was in line with the number in the assembled genome (Figure 1C), indicating that the assembled genome provides an accurate representation of the Burro1 landscape.

We used two methods to estimate the age of Burro1: sequence divergence, and phylogeny. Assuming a stable rate of base substitutions, quantification of the sequence divergence between the current copies and the consensus sequence gives a relative estimate of a transposon’s age compared to other transposons ⁴⁴. Divergence analysis revealed that a significant subset of Burro1 elements exhibited higher divergence from the consensus compared to other Ty3 elements, suggesting that it is one of the oldest Ty3 families in the planarian genome (Figure 1D) ⁴¹. We also investigated when in flatworm evolutionary history Burro1 was first introduced. We confirmed its previously reported presence in Girardia tigrina ⁴², and also found evidence of its expression in Schmidtea polychroa, Dugesia japonica, and Girardia dorotocephala (all members of the Dugesiidae), as well as in Bipalium kewense (Geoplanidae). We however failed to detect Burro1 in Macrostomum lignano (Macrostomidae; outgroup), or in any of the members of the Planariidae or Dendrocoelidae (Figure 1E). This suggests that Burro1 first arose within the Geoplanoidea, after they had split off from the Dendrocoelidae and the Planariidae (Figure 1E). The split between Dugesiidae and Geoplanidae is estimated to have occurred over 135 million years ago (MYA) ^45,46, thereby dating Burro1 as over 100 million years old.

Transposons typically go through life cycles ⁴⁷: initially, a new transposon rapidly expands to become established in the genome, resulting in the presence of many copies with high mutual similarity; later, activity is reduced, often leading to the death of the transposon ^48,49, which presents as transposons with fewer copies and high inter-copy divergence. To determine where Burro1 falls on this spectrum, we compared divergence and copy numbers of Ty3 transposons in the Schmidtea genome. Most elements followed the pattern described, showing either high copy numbers with low divergence or low copy numbers with high divergence (Figure 1F). Burro1 however exhibited both high mean divergence and high copy number. This suggests that the regulation of Burro1 is atypical, and that Burro1 has retained its activity.

To evaluate recent activity of the Burro transposons, we compared the 5' and 3' LTRs of full-length retrotransposon copies (Figure 1G). Because the two LTRs must have been identical at the time of insertion, and diverge over time ⁵⁰, copies with identical LTRs serve as markers of recent activity. Indeed, the ancient transposon Gypsy14 showed no copies with identical LTRs, whereas the currently active transposon SLF9 had many. For Burro1, while many elements had diverging LTRs, suggesting older insertions, 13% of the elements had LTRs with over 99% identity, providing strong evidence of recent activity (Figure 1G). Further, we serendipitously found at least one fully intact insertion of Burro1 that is present in the sexual strain of S. mediterranea and is absent from the asexual strain (Figure S1C). These strains are thought to have diverged around 1 MYA, indicating that Burro1 has been recently active.

Together, these data indicate that Burro1 is an ancient transposon that entered the flatworm genome over 100 MYA, after the split of Geoplanoidea from other planarian branches. Burro1 effectively expanded, and surprisingly shows signs of continued conservation and activity in the S. mediterranea genome.

Burro is composed of elements pre-existing in the planarian genome

To determine the origin of Burro1, we analysed the phylogenetic relations of its various parts to related elements in metazoan and viral genomes. The reverse transcriptase and integrase sequences of the three Burro elements were closely related to each other, and resembled those of other Ty3-related elements in the Schmidtea genome (Figure S1D). This suggests that the core Ty3 part of Burro likely came from inside the planarian genome rather than from horizontal gene transfer. The LTRs of the three Burro elements were not related, and each resembled LTRs from different planarian Ty3 elements (Figure S2A), indicating that the three Burro transposons likely originate from a single active Ty3 core element that acquired different LTRs (Figure S2C).

The most distinctive features of the Burro elements are the ORFs downstream of the polyprotein, and these also differ between the three Burro transposons. Several of the ORFs were small and showed no similarity to any known protein or nucleic acid sequences, but two ORFs in Burro1 contained recognizable protein domains, suggesting that they may have conserved functionality (Figure 1A). The first ORF downstream of the polyprotein contained a MATH domain and showed similarity to TNF-receptor associated proteins (TRAFs). The Burro-encoded ORF was similar between Burro1 and Burro2, and was closely related to other planarian MATH domain proteins (Figure S1A,S2B). This protein family is highly expanded in the planarian genome, encompassing well over 100 sequences ⁵¹. Their function has not been resolved as the planarian genome does not encode a clear homolog of TNF nor of the TNF receptor, and many of the TRAF-related genes lack the domain organization typical for TRAF proteins in other systems.

The second long downstream ORF of Burro1 encoded a protein with a well-conserved BIR domain and a RING domain, which is characteristic of the Baculoviral Inhibitor of Apoptosis (IAP) family ^52,53. Members of this protein family are found in eukaryotes, prokaryotes, and viruses, and are categorized into two different types. Proteins with type I BIR domains (which include vertebrate c-IAP, XIAP, and NAIP) typically exhibit caspase binding activity and function in the regulation of programmed cell death, whereas type II BIR domains are found in eukaryote proteins involved in the regulation of cell cycle (such as vertebrate proteins Bruce and Survivin). Viral BIR domains are diverse, suggesting that they do not have a monophyletic origin (Figure 2A).

Figure 2. Burro-iap (B-iap) is expressed in planarian neoblasts.

(A) Phylogenetic tree of metazoan BIR domains showing that Burro-IAP (B-IAP) is most closely related to other planarian type I BIR domain proteins. (B) Top: graph of the percentage of sequence similarity between the Burro1 sequences from various Dugeasiidae, showing sequence conservation in the core Ty3 domain as well as in the B-IAP region. Bottom: heatmap representing the rate of synonymous vs non-synonymous mutations (dS-dN) in the different coding regions of Burro1. (C) qPCR quantification of RNA associated with monosomes (untranslated, green) and polysomes (actively translated transcripts, magenta) shows enrichment of B-iap transcript on polysomes, suggesting active translation. Housekeeping genes are shown as positive controls whereas histone transcripts that are not actively translated outside S-phase are use as negative controls. Values from two independent polysome fractionations are shown. (D) Western blot showing presence of the B-IAP protein in the planarian lysate. (E) Heatmap representing z-scores of expression levels from isolated planarian tissues ⁵⁵. B-iap is enriched in the neoblasts compared to other tissues. (F) Fluorescent in situ hybridization (FISH) on whole mount animals showing B-iap or the genomic iap transcripts (magenta) together with the intestinal marker mat (green). B-iap is expressed in the parenchymal space. Bottom panels show zoomed in sections. Scale bars: 100 μm. (G) FISH on isolated cells shows co-expression of B-iap (magenta) and the stem cell marker smedwi-1 (green). Scale bar: 10 μm. (H) Immunostaining of intact planarians for the B-IAP protein (magenta) combined with FISH for the neoblast marker smedwi-1 (green) shows granules of B-IAP protein associated with the neoblasts. Scale bar: 10 μm. (I) Immunostaining of dissociated cells for the B-IAP protein (magenta) combined with FISH for the neoblast marker smedwi-1 (green) shows B-IAP protein in the neoblasts. Scale bar: 10 μm.

See also Supplemental Figure S3.

To determine the origin of the Burro1 BIR-domain protein, we performed a phylogenetic analysis. We identified 5 other BIR-domain proteins in the Schmidtea genome of which the two type I BIR domains (from IAP-1 and IAP-2, Figures 2A, S1E) clustered closely with the Burro1 BIR domain. We therefore refer to the Burro1-encoded protein as Burro-IAP (B-IAP). Interestingly, iap-1 and iap-2 had very similar exon structures and retained considerable sequence similarity. B-iap, which lacks introns, had 47.8% identity to iap-2, suggesting that it originated from a cDNA copy of this gene.

Taken together, these analyses suggest that Burro1 was formed by combining several elements that pre-existed in the planarian genome (Figures 1H, S2C). The downstream ORFs are closely related to their genomic genic counterparts, indicating that they likely are of planarian origin rather than acquired by horizontal gene transfer.

Burro-IAP (B-IAP) is a conserved protein and is expressed in planarian neoblasts

To identify functional elements in Burro1, we analysed its DNA sequence across planarian species to evaluate signs of evolutionary pressure. As expected, the core Ty3 protein region showed clear sequence conservation between different species (Figure 2B). Interestingly, we detected a second region of conservation, covering the downstream ORF that corresponds to B-IAP. As some changes in DNA sequence result in conservation of the amino acid sequence, we analyzed the ratio of synonymous to non-synonymous changes in the various elements of Burro1 (Figure 2B). While non-synonymous mutations were common in the 5' extended region of the polyprotein and in the 3' additional ORFs, the core Ty3 region and the B-iap region showed notably lower levels of amino acid substitutions. This indicates that the B-IAP protein is under selective pressure, suggesting that it is functionally important.

Active protein translation typically takes place on polysomes, whereas non-coding or aberrant transcripts are retained at monosomes and degraded. To determine whether B-iap was translated into protein, we used qPCR to analyze the distribution of the B-iap transcript between monosomes and polysomes (Figure 2C) ⁵⁴. Similar to coding genes, the B-iap transcript was enriched on polysomes, suggesting that its active translation. To identify the protein product, we raised and affinity-purified a B-IAP-specific antibody (Figure S3A). Application of this antibody to planarian lysates revealed a specific band around 50kD which was reduced upon the RNAi-mediated knockdown of Burro1 (Figure 2D), indicating that the Burro1-encoded B-iap sequence indeed produces a full-length protein product in planarian cells.

We inspected RNAseq data from various isolated tissues ⁵⁵ to determine the expression pattern of B-iap and to compare this to the two genomic iap genes. We found that iap-1 and iap-2 were broadly expressed at a low level, whereas B-iap was more highly expressed. iap-2 was enriched in epidermis and intestine, whereas iap-1 and B-iap were clearly detected in the adult pluripotent stem cells of planarians, known as the neoblasts (Figure 2E). This was confirmed by public single cell sequencing data (Figure S3B), and by qPCR analysis on isolated neoblasts (Figure S3C). We next used Fluorescent In Situ Hybridization (FISH) to independently verify the tissue distribution of B-iap (Figure 2F). In agreement with the RNAseq data, iap-2 was mostly present in subsets of epidermal and intestinal cells. The B-iap transcript was detected at higher intensity, and was enriched in parenchymal cells, which include the neoblasts. FISH on dissociated cells as well as on whole mounts confirmed that B-iap expression was detected in the same cells as the neoblast-specific marker smedwi-1 (Figures 2G, S3D). Using the B-IAP antibody we found that some of the B-IAP protein was localized to foci associated with the stem cells (Figures 2H, S3E). To specifically investigate the intracellular protein, we analyzed B-IAP in dissociated cells co-stained with markers for specific cell types. This showed that B-IAP protein was present in neoblasts (Figures 2I, S3F) and in intestinal cells (Figure S3G), but not in epidermal cells.

B-IAP is upregulated in response to stress

While exploring the expression of the planarian iap genes, we found that B-iap levels significantly increased immediately upon amputation, whereas the other iap transcripts were not affected (Figures 3A, S3H,I). This effect was confirmed by reanalysis of previously published sequencing data from a regeneration timecourse ⁵⁶ (Figure S3J). B-iap transcript was detected throughout the tissue fragment and notably included expression in the neoblasts (Figure 3B). Although levels of the Piwi transcript smedwi-1 were slightly decreased after amputation, suggesting a possible loss of transposon silencing, SMEDWI-1 protein levels remained unaffected. Furthermore, knockdown of smedwi-1 failed to increase burro1 or B-iap levels (Figure S3K-M).These results indicate that the amputation-induced increase in B-iap is not caused by a loss of SMEDWI-1-mediated silencing.

Figure 3. Burro-IAP (B-IAP) is upregulated in response to stress.

(A) qPCR on tissue fragments in the early hours upon head amputation (hpa) shows an increase of B-iap transcript at 1 hour after amputation. Data shown as mean ± SD; n=6 0hpa samples, and 4 1hpa samples of 3 animals each. (B) FISH at 1 hour after amputation shows an increase in B-iap transcript (magenta) throughout the tissue fragment. Scale bars: 50 μm. Insets show expression in neoblasts by co-localization with the neoblast marker smedwi-1 (green). (C) qPCR at 1h after amputation on tissue fragments treated with the transcription inhibitor ActinomycinD or DMSO as a control, shows an increase in the early wound-response transcript jun and in B-iap transcript only in the control-treated samples. Data shown as mean ± SD; n=5 independent samples of 3 animals each. (D) qPCR at 1h after irradiation (1500Rads) , drought, or heat stress shows an increase in B-iap transcript but no change in transcripts for smedwi-1, iap-1, or iap-2. Data shown as mean ± SD; n=5 independent samples of 3 animals each. (E) Western blot of planarian lysate at indicated time points after amputation shows an increase in B-IAP protein. Quantification relative to Tubulin signal. (F) Western blot of planarian lysate at indicated time points after irradiation shows an increase in B-IAP protein. Quantification relative to Tubulin signal.

See also Supplemental Figure S3.

To test whether the increase in B-iap was driven by increased transcription or decreased turnover, we repeated the amputations in the presence of the transcription inhibitor Actinomycin D (Figure 3C). As a control, we observed that activation of the early wound response gene jun was repressed by Actinomycin D, confirming effective inhibition of transcription. Under these conditions, the increase in B-iap RNA was completely abolished. These results indicate that the amputation-induced rise in B-iap levels is a result of increased transcription. This transcriptional response preceded the activation of mitosis in the neoblasts, or the increase of apoptosis detected at the wound site ^57,58, which occur at 6 hours and 4 hours after wounding respectively, emphasizing that B-iap is part of a very early response to injury. Of note, a second peak of B-iap expression was detected at 72 hours after amputation, concomitant with increase in the stem cell transcript smedwi-1 (Figure S3N,O), indicating that this late increase in B-iap could be the result of stem cell expansion.

To test whether the transcriptional upregulation of B-iap extended to other forms of organismal stress, we tested B-iap levels upon exposure to irradiation, drought, and heat stress. We found that at 1 hour after each of these stresses B-iap levels were strongly increased (Figures 3D, S3P), indicating that planarian cells rapidly upregulate B-iap transcription upon a wide range of organismal stresses. We further tested whether the detected upregulation of B-iap at the transcript level led to elevated B-IAP protein levels. Indeed, both amputation (Figure 3E) and irradiation (Figure 3F) resulted in notably increased levels of B-IAP protein as detected by Western blot.

Burro-IAP is a functional caspase inhibitor

IAP proteins are best known for their anti-apoptotic activity through direct interactions with caspase proteins ⁵⁹. The best studied mammalian IAP protein, XIAP, contains three BIR domains: second BIR domain binds to effector caspases (Caspase-3 and -7) resulting in blocked enzymatic activity, while the third binds the initiator caspase, Caspase-9, inhibiting its dimerization. In particular the obstruction of Caspase-9 dimerization is essential for XIAP’s anti-apoptotic effect ⁶⁰. Planarians encode both effector caspases and initiator caspases (Figure S4A,B). We thus hypothesized that B-IAP might interact with one or more planarian caspase proteins.

To determine whether B-IAP retained the amino acid residues required for caspase inhibition, we aligned its BIR domain with the well-characterized human BIR domains (Figure S4C). We found that the B-IAP lacked the residues required for enzymatic inhibition of effector caspases, but the residues for inhibition of the initiator caspases were largely retained ^61-63. Structural modeling of the B-IAP BIR domain with the planarian caspases, revealed a plausible interaction with one of the four initiator caspases whereas no such interaction was recovered for any of the effector caspases (Figures 4A, S4D,E). To further explore the binding of B-IAP to initiator caspases, we cloned tagged versions of both proteins into HEK293T cells to test their interaction in vitro (Figures 4B, S4F,G). B-IAP indeed co-precipitated with the predicted initiator caspase (Caspase_10708), but we found an even stronger interaction with the related planarian initiator Caspase_9038. We next used tagged Caspase_9038 protein to pull down interactors from planarian lysate, and indeed precipitated B-IAP protein (Figures 4C, S4H). These data demonstrate that B-IAP is a genuine caspase-binding protein.

Figure 4. Burro-IAP (B-IAP) is a functional anti-apoptotic protein.

(A) Heatmap showing the likelihood of interaction between the three planarian IAP proteins and the various planarian caspases (dd IDs (37)). ipTM: confidence score for the interaction. Only the interaction between B-IAP and the initiator caspase has significant support. For reference, ipTM for human XIAP with caspase9: 0.75 (B) Western blots showing coIPs of HA-tagged caspases with FLAG-tagged IAP protein expressed in HEK293T cells. The predicted interaction between B-IAP and caspase_10708 is confirmed, but a stronger interaction with caspase_9038 is identified. (C) Western blots showing coIPs of planarian B-IAP from planarian lysates with exogenously supplemented FLAG-tagged caspase_9038 protein. (D) Plots showing in vitro caspase activity in control samples and samples supplemented with the B-IAP protein. Data shown as mean ± SD. Significance: **p ≤ 0.01 by Student's t test. n=3 planarian lysates for each control and each supplemented datapoint.

See also Supplemental Figure S4.

Dimerization of human Caspase-9 triggers its proteolytic activity, which activates effector caspases to cleave a large set of cellular proteins and drive the cell into apoptosis. To further evaluate the anti-apoptotic capacities of B-IAP, we expressed the protein in HEK293T cells and tested its effect on initiator caspase activity in planarian lysates (Figure S4I). We determined that the consensus cleavage site in the planarian effector caspases consists of a VLAD motif (Figure S4J), and used in vitro fluorometric substrates with similar peptide sequences to quantify the effect of exogenous B-IAP protein on caspase activity. We found that the cleavage activity matching the peptide preference of the initiator caspase (but not other peptides) was significantly reduced (Figure 4D), indicating that B-IAP functions as a potent inhibitor of planarian initiator caspases.

Together, our data indicate that B-IAP binds planarian initiator caspases and blocks their cleavage activity, thereby accomplishing an anti-apoptotic effect.

Burro1 is required for normal neoblast function and regeneration

Given the presence of the anti-apoptotic B-IAP protein in Burro1, we hypothesized that Burro1 might have a positive effect on planarian physiology. We used RNAi-mediated knockdown to obtain a significant reduction of the B-iap transcript levels without affecting the genomic IAPs or other Burro transposons (Figure S5A,B). This resulted in a reduction in B-IAP protein (Figure 2D), and caused a moderate but significant reduction in the transcript levels of several stem cell (neoblast) genes (Figures 5A, S5A). FISH for the neoblast marker smedwi-1 (Figure 5B), and quantification of mitotic cells as labeled by phosphorylated histone 3 (H3P) (Figure 5C) confirmed this reduction in the number of neoblasts in Burro1(RNAi) animals.

Figure 5. Burro1 supports regeneration in S. mediterranea.

(A) qPCR quantification of transcript levels of several stem cell genes in Burro1(RNAi) samples compared to control RNAi. n=3 samples of 2 animals each. Data shown as mean ± SD. (B) Representative images of FISH on whole mount control animals and Burro1(RNAi) animals showing the stem cell marker smedwi-1 (green). Right: quantification of the images. Scale bar: 50 μm. n=3 individual animals per condition. (C) Representative images of immunostaining on whole mount control animals and Burro1(RNAi) animals showing the mitotic cell marker H3P (phosphorylated Histone 3; green). Right: quantification of the images. Scale bar: 500 μm. n=10 individual animals per condition. (D) Schematic of head regeneration experiment in control animals and Burro1(RNAi) animals at 6 days after amputation. (E) Representative live images of regenerating tail pieces of control animals and Burro1(RNAi) animals at indicated timepoints after amputation. Scale bar: 1 mm. n=8 individual animals per condition. (F) Representative images of FISH for anterior neuronal markers cintillio (magenta) and glutamate decarboxylase (green), and pharynx marker laminin in regenerating tail pieces of control animals and Burro1(RNAi) animals at indicated timepoints after amputation. Scale bar: 500 μm. n=5 individual animals for d6 timepoints; 4 animals for d14 controls, and 6 animals for d14 Burro1(RNAi). (G) Quantification of the neuronal markers shown in F. (H) Representative images of immunostaining on tail pieces of control animals and Burro1(RNAi) animals showing the mitotic cell marker H3P (green) 4 hours after amputation. Right: quantification of the images. Scale bar: 200 μm. n=7 individual animals per condition. (I) Representative image of TUNEL staining on tail pieces from control and Burro1(RNAi) animals 4 hours after amputation, showing increased apoptosis upon knockdown of Burro1. Right: quantification of the images. Scale bar: 200 μm. n=5 individual animals per condition. (J) Western blot showing the level of activated Caspase-3 in control animals and Burro1(RNAi) animals 4 hours after amputation. Quantification relative to Tubulin signal. (K) Representative image (tile scan) of TUNEL staining on control and Burro1(RNAi) animals 1 day after heat stress, showing increased apoptosis upon knockdown of Burro1. Right: quantification of the images. Scale bar: 300 μm. n=8 individual animals per condition. (L) Representative image (tile scan) of FISH staining for neoblast markers (bruli and h2b) on control and Burro1(RNAi) animals 3 days after 1250Rads of irradiation, showing reduced neoblast survival in the Burro1(RNAi) animals. Right: quantification of the images. Scale bar: 500 μm. n=6 individual animals per condition.

Significance A,B,C,G,H,I,K,L: *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, by Student's t test.

See also Supplemental Figure S5.

To assess the functional relevance of this neoblast reduction, we measured the regenerative response following head amputation. Head regeneration was significantly delayed in Burro1(RNAi) animals compared to controls (Figure 5D,E). FISH analysis of anterior neuronal and pharyngeal markers (cintillo, glutamate decarboxylase, and laminin respectively) revealed both fewer anterior neuronal cells and a less developed pharynx in the Burro1(RNAi) animals at 6 days post-amputation (Figure 5F,G). Despite this initial delay, regeneration eventually proceeded and by 14 days Burro1(RNAi) animals had largely completed the process.

The early response to tissue amputation typically involves localized apoptosis at the wound site, combined with a systemic increase in mitosis, followed 2 days later by increased local mitosis at the wound site to regenerate the missing tissue ^57,58. However, upon reduction of B-IAP, we observed subdued early mitotic activation as marked by H3P (Figure 5H), alongside a significant increase in apoptotic bodies as detected by TUNEL (Figure 5I). Apoptosis extended throughout the parenchymal region, where the neoblasts reside, suggesting that the increased apoptosis in the absence of B-IAP may have caused the observed regeneration delay. Interestingly, knocking down Caspase_9038 had the opposite effect, modestly accelerating regeneration. This was most evident at day 6 post-amputation when caspase_9038(RNAi) animals had a fully formed pharynx and mouth, whereas the pharynx of control animals remained underdeveloped and disconnected from the ventral surface (Figure S5C). Furthermore, knockdown of caspase_9038 largely rescued the regeneration phenotype of the Burro1(RNAi) animals (Figure S5D), indicating that the Burro1(RNAi) regeneration delay was likely driven by overactivity of the caspase. To further verify the impact of B-IAP on the level of apoptosis during regeneration, we quantified activated (cleaved) Caspase-3 upon amputation by Western blot (Figure 5J), and indeed detected a significant increase in the level of cleaved caspase, indicating an increase in apoptotic activity, in the absence of B-IAP.

We tested the physiological responses to several other stress conditions to assess whether the protective effect of Burro1 extended to these conditions. We found that heat stress resulted in a modest level of apoptotic cells in control animals, but in Burro1(RNAi) animals the density of apoptotic cells was significantly increased (Figures 5K, S5E). Further, we found that a dose of irradiation that consistently left a considerable number of neoblasts in controls, resulted in a significantly stronger loss of neoblasts in the Burro1(RNAi) animals (Figure 5L), whereas caspase(RNAi) animals exhibited the inverse effect (Figure S5F,G).

Together, our findings show that the active giant transposon Burro1 expresses a functional anti-apoptotic protein that protects planarian stem cells from apoptosis and enhances their regenerative abilities. This is a striking example of mutualism between a transposon and its eukaryote host, and provides an exciting model for how transposons can positively impact the fitness of their host.

Discussion

The interaction between transposons and their hosts has been described as an arms race, in which the host is best served by the inactivation or elimination of the transposon. While this is probably true for most transposons, we here describe a different collaborative strategy between transposon and host - one that supports maintenance of the transposon and benefits the host. We find that the giant retrotransposon Burro1 includes sequence encoding the BIR-domain protein Burro-IAP (B-IAP) that was derived from its host. B-IAP has retained its function as an anti-apoptotic protein that inhibits the activation of host caspases, but its expression has become stress-responsive. This provides stem cells carrying Burro1 with increased stress resistance and enhances host regeneration, albeit at the cost of higher transcriptional noise and the inherent risks of mobilization. It appears that this trade-off results in the positive selection of cells with Burro1, aiding the maintenance and expansion of this transposon.

Capture of non-transposon ORFs has previously been observed in retroviruses ⁶⁴ and in retrotransposons in rotifers and plants ^65-67. Some of these ORFs resemble endoviral ENV proteins and may enable horizontal spreading. Most accessory ORFs however lack similarity to any known genes, and their role in transposon survival or host interaction has thus remained unresolved. In the case of B-IAP however, recognizable domains pointed to a specific function in apoptosis, and we were therefore able to establish that this transposon-captured protein directly and impactfully alters the biology of its host cell.

Interestingly, transposable elements have been linked to the process of regeneration in several other highly regenerative animals. In hydra, axolotl, newt, and sea cucumber, transposons were upregulated upon injury ^68-71. An intriguing possibility is that a transient release of transposon silencing serves to prime defenses against related elements, as proposed for some transposon activation events during development ^72-75. In sea cucumber it was noted that although apoptosis around the wound site was prominent, the transposon-expressing cells survived and contributed to regenerated tissues, suggesting that transposon expression could confer a protective benefit ⁶⁹. And a recent earthworm genome assembly revealed that genes expressed during regeneration tended to be associated with transposable elements ⁷⁶. Together, these findings suggest the widespread activation of transposons upon wounding in regenerative animals, which may render them ideally positioned to contribute elements that enhance the cellular stress response.

The connection between retrotransposons and regulation of apoptosis has also been noted in several other contexts. Gene duplications introduced by retrotransposons resulted in the massive expansion of p53 gene copies in large mammals ⁷⁷. In gibbons, a retrotransposon was found to provide enhancers to several genes involved in DNA break repair ⁷⁸, and throughout mammalian genomes, IAP family genes have acquired enhancer elements from retrotransposons ⁷⁹. Finally, in oysters and clams, IAP elements have dramatically expanded in the genome, in part through the action of retrotransposons ^80,81. This raises the notion that transposons may have repeatedly evolved the ability to modulate the level of apoptosis in their hosts. It is possible that transcriptional noise or DNA damage induced by some transposons are so significant, that suppressing host apoptosis is an essential survival strategy for these elements themselves, although the modest activity of most extant elements may argue against this. Regardless, these findings make a strong argument for investigating whether other accessory ORFs with currently unknown homology also function as apoptotic modulators.

The notion that Burro1 may have enhanced the regenerative capabilities of planarians is enticing. While our lack of data regarding the regenerative abilities of the Burro-less ancestor makes it difficult to prove this definitively, the distribution of regenerative abilities among extant planarians provides some support. Most planarians can regenerate, yet the Dugesiidae (which is the main clade carrying the Burro transposon) exhibit substantially improved regenerative abilities compared to other planarian families ⁸². Closer inspection of Dugesiidae transcriptomes revealed that all contain a Burro-related Ty3 sequence, and multiple copies of iap-2 / B-iap. Although B-iap and iap-2 sequences are difficult to distinguish phylogenetically, the cooccurrence of a Burro-like Ty3 sequence with multiple copies of the iap-2 / B-iap sequence makes it highly probable that Burro1 is present and transcribed throughout this clade. Interestingly, Cura pinguis, the only species in this clade with more limited regeneration, contains a Burro-like Ty3 element, but lacks a second iap-2 like sequence, suggesting that its Burro1 may be incomplete. In contrast, the Planariidae clade, which exhibited intermediate regenerative abilities, lacks Burro-related sequences entirely. However, the robustly regenerating species in this group ⁸² all encode multiple copies of the iap-1 gene, a unique feature of this clade. While there certainly are many other differences in genomic makeup between these planarian clades, these observations raise the possibility that expanding the iap copy enhances regeneration and that expansion of Burro1 was an effective way to achieve this in Dugesiidae.

Transposon mutualisms are probably rare, and they may be prone to dissolve into parasitism by loss of the benefit that the transposon provides, or into domestication by bringing the beneficial element under host control ^83,84. There may however be a benefit to retaining B-iap on an active mobile element rather than as a domesticated gene. In the first place, transposable elements have long been proposed to activate under stress ^85-87. The presence of the anti-apoptotic B-iap on a transposon thus ensures the activation of the gene when it is most needed without the need to recruit a specific promoter. In the second place, under persistent stress, the lasting elevated expression of Burro1 will likely result in new Burro1 insertions, thereby dynamically increasing the B-iap gene dosage and the stress resistance of the host cells when the environmental conditions demand it, even generating heterogeneity among the existing cells to optimize the dose. This may be particularly important for long-lived cells, such as the planarian neoblasts, that thus acquire an additional mechanism of adaptation to changing conditions.

The existence of beneficial interactions with transposons may also impact the evolution of host defense mechanisms. The primary eukaryotic defenses (piRNAs, siRNAs, KRAB-ZNFs) are sequence-specific, and thus target each transposon individually. This specificity may allow the host to titrate its defensive response based on the benefit provided by each particular element. By uncovering mutualistic features within a eukaryote mobile element, our study reveals an unexpected alliance between transposon and host, and enhances our appreciation of the intricate interplay between transposons, host defense systems, and the trajectory of genome evolution.

Limitations of the study

To study the effect of the B-IAP protein on planarian physiology, we used RNAi to knock down the Burro1 transcript. This results in a reduction of the B-IAP protein level, but residual protein remains. The effect of full elimination of B-IAP therefore may be stronger than the effects reported in this study. Further, because B-iap is part of the Burro1 transcript we are unable to separate effects of knockdown of the B-IAP protein from potential effects from the rest of the Burro1 transposon, and the effect of the presence of the B-iap sequence on Burro1 propagation efficiency cannot be explicitly tested.

Models regarding the origin of Burro1 and the evolutionary effect of Burro1 in platyhelminths are based on data of extant species, and thus are inferences only.

STAR Methods

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Planarian strain and husbandry

Schmidtea mediterranea asexual clonal strain ClW4 was maintained as previously described ⁹⁰. Briefly, animals were cultured in 1x Montjuic salts at 20°C, fed homogenized beef liver paste every 1–2 weeks, and expanded through continuous cycles of amputation or fissioning and regeneration. Animals were starved 1-2 weeks prior to experiments, and assigned to control or treatment by randomized selection. For RNAi experiments, water was supplemented with 50ug/ml Gentamicin sulfate (VWR) to prevent bacterial growth.

Cell culture

Human embryonic kidney 293 (HEK293) cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM; high glucose) with 10% fetal bovine serum and 1% penicillin-streptomycin. Cultures were incubated at 37°C in a humidified atmosphere containing 5% CO₂. Cells were routinely passaged at ~80% confluence using trypsin-EDTA.

HEK293 cells were originally derived from the kidney of a female human embryo. Cells were obtained directly from ATCC where authentication is conducted prior to distribution.

METHOD DETAILS

RNAi and drug treatment

Regions of planarian genes 0.5-2kb in length were amplified from complementary DNA (cDNA) using sequence specific primers (Table S1). The PCR product was cloned into the pGEM-T vector (Promega) and verified by Sanger sequencing. Both RNA strands were synthesized in vitro from PCR-generated forward and reverse templates with flanking T7 promoters (TAATACGACTCACTATAGG), and annealed by incubation at 37°C for 30min. The transcribed ssRNAs as well as the final hybridized dsRNA product were verified by gel electrophoresis.

Size-matched animals were starved 1-2 weeks prior to experiments and assigned to treatment conditions by random selection. Experiments were performed on batches of 6-12 animals, and were repeated at least 3 times. No technical replicates are used.

RNAi food was prepared by combining 2ul dsRNA with 2ul food coloring and 50ul of homogenized beef liver ⁹¹. Burro1(RNAi) animals were fed dsRNA on day 0, and subsequently injected with dsRNA (without additions) three more times (on day 2, day 7 and day 9). DsRNA matching C. elegans gene unc-22 was used as a negative control. Efficiency of RNAi was verified by qPCR prior to further experiments, and if necessary, animals with ineffective RNAi were removed.

The RNAi constructs used in this study targeted the B-iap part of the Burro1 transposon. However, as Burro1 is likely mainly transcribed as one long transcript that originates from the 5’LTR, targeting of B-iap results in simultaneous reduction of the rest of the Burro1 transcript. For inhibition of transcription, 2ul of Actinomycin D (5mg/ml in DMSO) was combined with 2ul food coloring and 35ul of homogenized beef liver. Animals were processed for amputation at 16 hours after feeding. DMSO was used instead of Actinomycin as a control treatment.

Stress conditions

Amputations were performed posterior to the pharynx by razor blade incision. For irradiation, animals were placed in 6cm dishes with 6ml of water and exposed to the indicated dose of gamma irradiation after Thoreaux filter in a MultiRad350 (Precision Inc) in exposure control mode. Animals were subsequently maintained in water supplemented with 50ug/ml Gentamicin sulfate (VWR).

For drought exposure, animals were transferred to a clean 6 cm dish. Water was removed carefully, avoiding the collection of animals at the edges of the plate. Animals were left without water for 20 minutes, after which water was replenished.

For heat stress, animals were moved to a clean plate and water was replaced by planarian water preheated to 30°C. Plates were incubated in an oven at 30°C for 10 min. After the incubation was complete, water was removed and replaced by room temperature water three times to rapidly bring the temperature down.

Whole-mount fluorescent in situ hybridization and immunofluorescence

Fixations and whole-mount in situ hybridizations (ISH) and immunofluorescence were performed as previously described ⁹², with alterations described in ⁹³. Briefly, formaldehyde fixed animals were bleached using formamide bleach solution and treated with proteinase K (2 ug/ml) in PBSTx (PBS containing 0.1% Triton X-100). For FISH, following overnight hybridization at 56°C, samples were washed sequentially in pre-hyb solution, 1:1 pre-hyb-2x SSC, 2x SSC and 0.2x SSC at 56°C. Probes were detected with anti-DIG POD (Roche 11207733910), anti-Fl-POD (Roche 11426346910), or anti-DNP-HRP (Perkin Elmer PF1129). After tyramide development ⁹³, peroxidase was inactivated by incubation in 1% sodium azide for 1.5 hour if additional tyramides were used. Specimens were counterstained with DAPI (Sigma). For immunofluorescence of phosphorylated histone 3, animals were blocked and incubated with primary antibody anti-phospho-Histone3[Ser10] (Millipore, clone 63-1C-8) 1:750 overnight, followed by incubation with goat anti-Mouse IgG HRP Conjugate (Life Technologies). Signals were developed using Tyramide SuperBoost^™ Kits (Invitrogen).

For TUNEL staining, after fixation and bleaching, animals were incubated in TUNEL reaction buffer (25mM Tris-HCl, 200 mM Sodium Cacodylate, 0.25 mg/ml BSA, 1mM Cobalt Chloride) for 30 min before addition of 8U of TdT enzyme (NEB) and 120pmol DIG-dUTP (Roche). Reactions were incubated for 1 hour at 37°C in a humid chamber, followed by washes in PBSTx and detection using anti-DIG-POD and tyramide amplification as described above.

Neoblast isolation and staining

Neoblasts in G2/M phase (X1) and differentiated cells (Xins) were isolated by Fluorescence-Activated Cell Sorting based on DNA content as determined by Hoechst intensity ⁹⁴, following standard procedures described previously ⁹⁵.

For staining of the isolated cells, FACS-sorted cells were washed in CMF, spotted onto poly-D-lysine coated coverslips (BD Biosciences) positioned in 24 well plates, allowed to settle for 30 minutes, and fixed in 4% PFA (in PBS) for 20 minutes at room temperature. Controls and treatment were always spotted on the same cover slip, and went through all staining steps in the same well. FISH labelings were carried out similarly to the whole-mount protocol, with wash steps and antibody incubations shortened to 10 minutes and 1 hour, respectively.

Microscopy and image analysis

Images were acquired on a Zeiss LSM800 Confocal Microscope. Control and RNAi animals were imaged with the same magnification, laser intensity and gain, at comparable anatomical position. Images were generated in Image J ⁹⁶. Cell counting and quantification of fluorescence intensity were performed using the automated quantification scripts in the Planarian Image Quantification (PIQ) package v1.0.5 ⁵⁴.

qPCR analysis

Total RNA was isolated by Trizol and quantified by Qubit. RNA was treated with DNaseI (Promega) for 30 minutes at 37°C and cleaned by precipitation on Ampure beads. cDNA was synthesized using oligo dT primers and ProtoScriptII (NEB) according to the manufacturer instructions. qPCR reactions were performed using EvaGreen mastermix (Biotium). Primers are listed in Supplementary Table S1. RT and qPCR reactions of samples and controls were run in parallel in the same plates. qPCRs were run on a QuantStudio 3 instrument (ABI) with the following program: 95°C, 20s; 40 cycles of 95°C, 5s; 60°C, 20s; followed by a melting curve analysis.

For the quantification of transposon copy numbers by genomic qPCR, five regions corresponding to four transposable elements—Slf9, Gy8, B2, B1 5'Math, and B1 IAP—were PCR-amplified from cDNA using sequence-specific primers (Table S1), and cloned into the pGEM-T vector (Promega). Positive clones were verified by Sanger sequencing. Plasmids (1 μg) were linearized with ScaI (NEB) at 37°C for 1 h, followed by enzyme inactivation at 80°C for 20 min. The number of molecules was calculated for each preparation, based on plasmid size and DNA concentration. Serial dilutions of each plasmid were prepared to generate standard curves with final copy numbers of 1×10⁷, 1×10⁶, 1×10⁵, 1×10⁴, and 1×10³ molecules/μL.

Genomic DNA was extracted from sexual and asexual Schmidtea mediterranea using the CTAB method ⁴¹. Residual RNA was removed by treatment with 1 μL RNase A (NEB) at 37°C for 30 min. DNA was further purified using NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel) per the manufacturer’s protocol. DNA quality and concentration were assessed using a Nanodrop spectrophotometer (DeNovix). Extracted genomic DNA from planarians was normalized to 0.1 ng/μL and used as a template for qPCR. Reactions were performed using a EvaGreen master mix (Biotium) with primers specific to each transposon target region (Table S1). Standard curves were run in parallel to genomic DNA samples, and the resulting Ct values were used to interpolate the absolute copy numbers of each transposon in genomic DNA samples.

To estimate the genomic copy number of each transposon, qPCR amplicon sequences were queried against the S. mediterranea reference genome (smed_chr_ref_v1.fa; ⁹⁷) using NCBI BLAST+ (default parameters). The number of BLAST hits corresponding to high-identity matches was counted as the bioinformatically predicted genomic copy number. To compare experimental copy numbers obtained by qPCR with BLAST-based predictions, values were normalized to the copy number of Slf9, which served as an internal reference due to its recent insertion and presumed low sequence divergence. Relative abundance for each transposon was calculated by dividing both qPCR-derived and BLAST-derived copy numbers by that of Slf9.

Polysome profiling

Polysome profiling was carried out as described in ⁹⁸ and ⁵⁴. Briefly, worms were treated with cyclohexamide (100 μg/ml) for 48 hours and lysates were prepared by flash-freezing the worms in liquid nitrogen using pre-chilled polysome extraction buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1.5 mM MgCl₂, 0.6% Triton X-100, 0.5 mM DTT, 1× protease inhibitor, and 100 μg/ml cycloheximide), followed by mechanical grinding and douncing. The lysates were cleared by centrifugation (12000 g for 10 min at 4°C). Cleared lysate was layered on the top of a 10–50% sucrose density gradient prepared in the buffer and subjected to ultracentrifugation at 40000 RPM for 16 h at 4°C (Beckman SW 41 Ti rotor). Fractions were collected by upward displacement and the absorbance was monitored at 254 nm to localize the monosome and polysome peaks. Monosome- and polysome-containing fractions were pooled, and RNA was purified by TRIzol extraction for cDNA synthesis.

Nanopore sequencing and 5’RACE

Nanopore sequencing libraries were prepared using the cDNA-PCR Sequencing Kit (SQK-PCS109, ONT) according to the manufacturer’s instructions. Additional RACE libraries were generated by reverse transcription and template-switching by Maxima H Minus Reverse Transcriptase (Thermo Fisher Scientific Inc.), using six reverse primers specific to Burro1 and a common adapter. Nanopore amplicons were generated from cDNA with LongAmp Taq master mix (New England Biolabs, USA) and the Nanopore cDNA sequencing kit according to the manufacturer’s instructions. PCR products were purified using the AMPure XP beads (Beckman Coulter, USA). Sequencing was performed on a MinION Mk1B system (Oxford Nanopore Technologies Ltd., ONT), using a FLO-MIN106 flow cell. Basecalling was performed using guppy (ONT guppy v4.4.1., github.com/nanoporetech/rerio). Adapters were trimmed using Porechop (github.com/rrwick/Porechop). Only reads with a minimum Q score of 10 were selected for analysis of Burro1 transcripts.

In vitro caspase assays

B-IAP was cloned before a FLAG tag in the pcDNA 3.1 vector and expressed in HEK293T cells by polyethyleneimine (PEI)-mediated transfection. Empty vector transfected into HEK293T was used as a negative control. HEK293T cells were lysed by pipetting in pre-chilled lysis buffer (20mM Tris-HCl pH7.5, 150mM NaCl, 1.5mM MgCl2, 0.6% Triton X-100, 0.5mM DTT, 1X protease inhibitor, 1 mM PMSF) and lysates were cleared by centrifugation at 5,000 × g for 5 min at 4°C.

Apoptosis was induced by incubating live animals in 1mM enoxacin water for 24 hours followed by exposure to 550mJ/cm2 of UV irradiation in a UV crosslinker (UVP) ^99-101. Two hours after UV exposure lysates were prepared by grinding the animals in pre-chilled lysis buffer (20mM Tris-HCl pH7.5, 150mM NaCl, 1.5mM MgCl2, 0.6% Triton X-100, 0.5mM DTT, 1X protease inhibitor, 1 mM PMSF). Lysates were subsequently cleared by centrifugation at 5,000 × g for 5 min at 4°C. Cleared planarian lysates were supplemented with a fluorescent substrate and HEK293T cell lysate (control, or containing B-IAP protein) and incubated for one hour at 37°C to measure caspase activity. We had determined that planarian initiator caspases cleave the VLAD motif, however as this motif is not favored by any mammalian caspase, no commercial Ac-VLAD-AFC was available. We thus tested several substrates for motifs with similar animo acid characteristics, namely Ac-YVAD-AFC (Cayman 17591), Ac-IETD-AFC (Cayman 17480), and Ac-LEHD-AFC (Cayman 17051), and Abcam Caspase 3 Detection Kit (ab102491) containing Ac-DEVD-AFC. Caspase activity was quantified using a plate reader (Bio-Tek; ex 400 nm / em 505 nm).

Protein structure predictions

Predictions of protein structure were performed using Alphafold2 in ColabFold ^102-104. BIR domains of the three planarian IAP proteins (IAP-1 residues 54-156, IAP-2 residues 83-185, BURRO1 IAP residues 62-164) and the eight planarians caspases (dd_Smed_v4_10708 residues 169-427, dd_Smed_v4_9038 residues 172-427, dd_Smed_v4_11193 residues 116-391, dd_Smed_v4_10883 residues 239-497, dd_Smed_v4_3121 residues 1-266, dd_Smed_v4_4986 residues 1-263, dd_Smed_v4_8326 residues 1-248, dd_Smed_v6_1167 residues 1-252) were used to predict interactions between the IAP proteins and the caspases. To compute the complex structure of the BIR domain with the caspase we used three cycles and 5 models, using the “multimer-2” option.

Predictions were rerun with Alphafold3 with similar outcomes.

In vitro interaction assays

HEK293 cells were maintained in Dulbecco’s modified Eagle’s medium (DMEM; Gibco) supplemented with 10% (v/v) fetal bovine serum (FBS; Gibco) and 1× penicillin–streptomycin (Gibco). Cells were seeded in 15 cm dishes to ~70% confluence and transiently transfected with a plasmid encoding C-terminally FLAG-tagged planarian B-IAP, C-terminally FLAG-tagged Genomic IAP-2, and C-terminally HA-tagged Caspase, using Polyethylenimine, Linear, MW 25000, Transfection Grade (PEI 25K^™) according to the manufacturer’s instructions. 48 hrs post-transfection, cells were washed twice with cold PBS and lysed in ice-cold lysis buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% Triton X-100, and 1× protease inhibitor cocktail (Roche)) for 5,000 × g for 5 min at 4°C, and the supernatants were incubated with either anti-FLAG (M2) magnetic agarose resin (Millipore Sigma) or protein G magnetic beads (NEB) for 2 hr 4°C with gentle rotation. Resin was washed three times with wash buffer (50 mM Tris-HCl pH 7.5, 150 mM NaCl, 0.1% Triton X-100), and bound protein was eluted with acid-glycine buffer. The eluates were denatured in protein loading buffer (60mM Tris-Cl pH6.8, 5% Glycerol, 1% SDS and 2.5% β-mercaptoethanol) and used for immunoblotting as described below.

To test the binding of native B-IAP protein to planarian caspase, C-terminally FLAG-tagged Caspase-9038 protein was generated in HEK293T cells as described above. HEK293T cell lysate was combined with planarian lysate generated in the same buffer and incubated with anti-FLAG (M2) magnetic agarose resin (Millipore Sigma) for 2 hours at 4°C. Resin was washed in wash buffer and bound protein was eluted with acid-glycine buffer. The eluates were denatured in protein loading buffer and used for immunoblotting as described below.

While both the computational predications and the in vitro experiments supported the interaction of the B-IAP protein with initiator caspases, the two methods produced slightly different outcomes. The computational modeling predicted an interaction of B-IAP with Caspase_10708. The in vitro experiment confirmed this interaction, but found a stronger interaction with Caspase_9038. Further, whereas the modeling had indicated that B-IAP was a better binder of caspases, in vitro experiments found that the genomic IAP-2 protein was similar in its binding efficiency. The reason for these discrepancies probably lies in two major limitations in the modelling. In the first place, the modelling used only the BIR domain of B-IAP, as modeling of the full IAP protein did not result in a reliable structure. The IP however used the full B-IAP protein. In the second place, caspases are typically cleaved into a small and large subunit preceded by a less structured N-terminal region. The modeling used the caspase without this N-terminal sequence because this region disrupted the predicted caspase structures. This region again was present in the in vitro experiments.

B-IAP antibody generation

GST-tagged B-IAP was cloned into pGEX. The recombinant protein constructs were overexpressed in BL21(DE3) by induction with 0.2 mM isopropyl thio-β-D-galactoside (IPTG) at 15°C and purified using pre-equilibrated glutathione-sepharose beads (Invitrogen) as described previously ⁵⁴. Buffers were supplemented with protease inhibitor cocktail (Roche), DNase I and RNase A to rule out interactions through nucleic acids. All the procedures were carried out at 4°C. Proteins were concentrated with membrane filters, snap frozen and stored in −80°C.

Immunizations of rabbit hosts were performed by Cocalico Biologicals, Inc. B-IAP antibodies were purified from rabbit serum by affinity purification using B-IAP protein coupled to CNBr-activated Sepharose (Cytiva). Purified antibodies were snap frozen and stored in −80°C.

SDS-PAGE and Western blotting

Individual 1-3mm sized animals were homogenized in protein loading buffer (60mM Tris-Cl pH6.8, 5% Glycerol, 1% SDS and 2.5% β-mercaptoethanol) and separated on 8% denaturing polyacrylamide gel. Samples were transferred to PVDF membrane, blocked in PBST (PBS with 0.1% Tween-20) with 3% milk powder, and incubated with the primary antibody followed by secondary antibody, in PBST with 1.5% milk powder. The following commercial antibodies were used: mouse anti-α-tubulin (Millipore) at 1:10000; caspase-3 (Abcam) 1:500; mouse anti-FLAG M2 antibody (1:5000); mouse Normal IgG (Cell Signaling Technologies); goat anti-Rabbit IgG HRP Conjugate at 1:10000 and goat anti-Mouse IgG HRP Conjugate (Life Technologies) at 1:10000. Additionally, B-IAP antibody (described above) at 1:1000, and antibodies labeling PIWI proteins SMEDWI-1and SMEDWI-2 ^54,55 at 1:1000 and 1:2000 respectively were used.

Phylogenetic analysis

Genome and transcriptome sequences were retrieved from publicly available databases including giri REPBASE (https://www.girinst.org/repbase/), planarian.jp (http://www.planarian.jp), WormBase Parasite (https://parasite.wormbase.org/ftp.html), NCBI GenBank (https://www.ncbi.nlm.nih.gov), SMART (https://smart.embl.de/domains.cgi), Zenodo (https://doi.org/10.5281/zenodo.8301321), and PlanMine (https://planmine.mpinat.mpg.de/planmine/begin.do). Accession numbers and dataset versions are detailed in Supplementary Table S2.

For the analysis of the transposon copies, consensus transposon sequences as listed on Repbase (https://www.girinst.org/repbase/) were used to identify individual copies in the S. mediterranea genome using RepeatMasker (v4.1.2-p1-foss-2020b) ¹⁰⁵.

For analysis of sequence variation in the I-region (sequence without the LTRs) of the Ty3 transposable elements (TE), genomic sequences corresponding to 90–110% of the I-region length—based on the Ty3 reference TE—were extracted from Ty3 elements mapped to the S. mediterranea genome and exported in FASTA format. TE sequences were grouped by subfamily and aligned separately using MAFFT v7 ¹⁰⁶. Pairwise evolutionary distances (p-distance) were calculated for each TE using MEGA11 ¹⁰⁷.

For analysis of the LTR divergence, only SLF, gy14, B1, and B2 TEs with both 5′ and 3′ long terminal repeats (LTRs) were retained for further analysis. The flanking 5′ and 3′ LTR sequences for each TE were extracted from the genome using SeqKit (v2.8.1) ¹⁰⁸ and exported in FASTA format. Pairwise global alignment between the 5′ and 3′ LTR sequences of each TE was performed using EMBOSS Needle ¹⁰⁹ to calculate percentage similarity.

Domain-containing sequences (IAP, LTR, MATH, and Ty3 pol) were identified using BLAST+ version 2.15.0 ¹¹⁰ on the Yale McCleary HPC cluster. Searches used blastn against genome and transcriptome assemblies with the -task blastn parameter and an E-value threshold of 1e-5. Query sequences were provided in unmasked FASTA format. Nucleotide sequences for each genic element were aligned using MAFFT version 7 ¹⁰⁶. Maximum likelihood (ML) analyses were performed with RAxML-NG v1.2.1 ¹¹¹ applying the GTR+G model of nucleotide substitution, with node support assessed by 20,000 bootstrap replicates. Bayesian phylogenetic inference was performed separately for each gene alignment using MRBAYES version 3.2.7 ¹¹². Two independent Markov chain Monte Carlo (MCMC) runs, each with four incrementally heated chains, were run for 10 million generations, sampling every 5,000 generations. The initial 25% of samples were discarded as burn-in, and convergence was confirmed by standard diagnostics. Consensus trees were constructed from post-burn-in samples. Phylogenetic trees were visualized and annotated with the Interactive Tree of Life (iTOL) online tool ¹¹³. Bootstrap and posterior probability support values were combined and annotated on the ML consensus tree using Adobe Illustrator.

Protein sequences of Caspases and BIR-domain proteins were aligned using MAFFT v7¹⁰⁶. Maximum likelihood analyses were conducted using RAxML-NG v1.2.1 ¹¹¹ with the LG+G8+F substitution model. The analysis was run using the --all option on the MAFFT-aligned dataset, with 2,000 bootstrap replicates (--bs-trees 2000), a bootstrap support cutoff of 0.01 (--bs-cutoff 0.01), and both FBP and TBE bootstrap metrics enabled (--bs-metric fbp,tbe). Analyses were performed using automatic thread allocation (--threads auto). Resulting ML trees were visualized using standard tree-viewing software (iTOL).

For the analysis of synonymous and non-synonymous mutations, the genome sequence of Dugesia japonica was obtained from Planarian.jp (http://www.planarian.jp) ¹¹⁴, and the transcriptome data were retrieved from PlanMine (https://planmine.mpinat.mpg.de). The Schmidtea mediterranea BURRO1_I sequence was used as a query for BLAST searches to identify the homologous BURRO1 sequence in D. japonica. Conserved regions between species were visualized using global alignments and VISTA ¹¹⁵. Synonymous (dS) and nonsynonymous (dN) substitution rates were calculated using MEGA11 ¹⁰⁷.

Processing of mRNA-seq data

Publicly available mRNA libraries (PRJNA276084) were reanalyzed by mapping against the Schmidtea mediterranea genome ¹¹⁶. Reads were filtered and trimmed using fastp (version 0.21.0)¹¹⁷ with parameters --length_required 20 --average_qual 20. Mapping was performed using STAR (version 2.7.2a) ¹¹⁸ with the following settings: --outFilterMismatchNmax 2 -- alignIntronMax 15000 --alignMatesGapMax 15000 --outFilterMultimapNmax 100 -- winAnchorMultimapNmax 100. FeatureCounts (version 1.6.4) ¹¹⁹ was used to count the reads/fragments over the gene annotations with parameters -M -C -O. Heatmaps were generated using pheatmap on R v4.10.

QUANTIFICATION AND STATISTICAL ANALYSIS

All reported measurements were made from distinct samples. Levels of significance were calculated by two-tailed Student’s t test, using the Prism software package. Error bars indicate standard deviations. Whiskers on box plots indicate min and max observations.

Supplementary Material

2

Table S2. Sequences used in the phylogenetic analysis. Related to STAR Methods.

(A) Bir domain sequences used for the phylogenetic analysis shown in Figure 2A (B) MATH domain sequences used for the phylogenetic analysis shown in Figure S2B. (C) Ty3-related polymerase sequences used for the phylogenetic analysis shown in Figure S1D. (D) LTR sequences used for the phylogenetic analysis shown in Figure S2A. (E) IAP sequences used for the phylogenetic analysis shown in Figure S1E.

Document S1. Figures S1–S5, Table S1, and supplemental references Table S2. Sequences used in the phylogenetic analysis.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit polyclonal B-IAP	This paper	N/A
Mouse monoclonal anti-FLAG	Sigma-Aldrich	Cat#F1804
Rabbit polyclonal anti-HA	Abcam	Cat#ab9110
Rabbit polyclonal anti-caspase-3	Abcam	Cat#ab13847
Mouse anti-α-tubulin	Millipore	Cat#MABT205
Mouse anti-actin	Millipore	Cat#MABT825
Mouse monoclonal anti-Rabbit IgG, HRP	Cell Signaling Technology	Cat#5127S
Rat monoclonal anti-mouse IgG, HRP	Abcam	Cat#ab131368
Mouse Normal IgG	Cell Signaling Technology	Cat#5415S
Goat polyclonal anti-Rabbit IgG, HRP	Abcam	Cat#ab6721
Goat polyclonal anti-Mouse IgG, HRP	Abcam	Cat#ab6789
Sheep polyclonal anti-DIG-POD	Roche	Cat#11207733910
Sheep polyclonal anti-Fl-POD	Roche	Cat#11426346910
Rat monoclonal anti-DNP-HRP	Perkin Elmer	Cat#PF1129
Bacterial and virus strains
DH10B F-mcrA Δ(mrr-hsdRMS-mcrBC) Φ80dlacZΔM15 ΔlacX74 endA1 recA1 deoR Δ(ara,leu)7697 araD139 galU galK nupG rpsL λ–	New England Biolabs	Cat#C3019I
BL12 (DE3) fhuA2 [lon] ompT gal (λ DE3) [dcm] ΔhsdS λ DE3 = λ sBamHIo ΔEcoRI-B int::(lacI::PlacUV5::T7 gene1) i21 Δnin5	New England Biolabs	Cat#C2530H
Chemicals, peptides, and recombinant proteins
Actinomycin-D (Act-D)	Sigma-Aldrich	Cat#A1410
Ampure XP beads	Beckman Coulter	Cat#A63881
Anti-FLAG® M2 Magnetic Beads	Sigma-Aldrich	Cat#M8823
Benzamidine Separopore 6B	bioWORLD	Cat#20181111-2
c0mplete Mini, EDTA-free	Roche	Cat#11836170001
cyclohexamide	Sigma-Aldrich	Cat#1810
Digoxigenin-11-dUTP	Roche	Cat#11093088910
DMEM, high glucose, pyruvate	GIBCO	Cat#11995065
DMSO	New England Biolabs	Cat#B0515A
DNaseI	New England Biolabs	Cat#M0303S
DPBS	GIBCO	Cat#14190-144
DTT	ThermoFisher	Cat#BP17225
Enoxacin	ThermoFisher	Cat#J61912.06
EvaGreen	New England Biolabs	Cat#1725213
Fetal bovine serum	Sigma-Aldrich	Cat#F0926
Formaldehyde	ThermoFisher	Cat#119690010
Formamide (deionized)	Ambion	Cat#AM9344
Gentamicin Sulfate Salt	Sigma-Aldrich	Cat#G1264-1G
glutathione-sepharose beads	ThermoFisher	Cat#16101
Hydrogen Peroxide	Sigma-Aldrich	Cat#7722-44-1
L-glutamine	GIBCO	Cat#25030-081
LongAmp Taq 2XMaster Mix	New England Biolabs	Cat#M0287
Maxima H Minus Reverse Transcriptase	ThermoFisher	Cat#EP0752
2-Mercaptoethanol	Sigma-Aldrich	Cat#M3148
Pen-strep	GIBCO	Cat#15140-122
PMSF	MP Biomedicals	Cat#219538105
Polyethylenimine (PEI)	Kyfora Bio	Cat#23966-100
Proteinase K	Invitrogen	Cat#25530049
Protoscript II RT	New England Biolabs	Cat#M0368L
RNaseH	New England Biolabs	Cat#M0297L
Scal-HF	New England Biolabs	Cat#R3122S
Terminal Transferase	New England Biolabs	Cat#M0315S
Triton X-100	Sigma-Aldrich	Cat#SLBW6852
Trizol	Life Technologies (ambion)	Cat#15596018
Trypsin-EDTA	Invitrogen	Cat#25300-054
Tween-20	Sigma-Aldrich	Cat#P7949
Critical commercial assays
Ac-YVAD-AFC	Cayman	Cat#17591
Ac-IETD-AFC	Cayman	Cat#17480
Ac-LEHD-AFC	Cayman	Cat#17051
Caspase 3 Detection Kit	Abcam	Cat#ab102491
cDNA-PCR sequencing Kit	Oxford Nanopore	Cat#SQK-PCS109
Monarch® Plasmid Miniprep Kit	New England Biolabs	Cat#T1110L
Nucleospin Gel and PCR Clean-up kit	Machery-Nagel	Cat#740609
Deposited data
Regeneration data re-processed (Fig S2I and S3E)	Wurtzel et al. 2015 56	PRJNA276084
Planarian transcriptomic data re-processed (Discussion)	Vila-Farre et al. 82	PRJNA1011852
Nanopore sequencing data	This paper	PRJNA1135975
Schmidtea mediterranea genome	Ivankovic et al.116	Planmine
Dugesia japonica genome	An et al. 2018 114	http://www.planarian.jp
Planarian transcriptomic data re-processed (Discussion)	Vila-Farre et al. 82	https://doi.org/10.5281/zenodo.8301321
Genome and Transcriptomes	This paper	see Supplementary Table S2
Western blots	This paper	Mendeley: 10.17632/8gk5ssdtvp.1
Experimental models: Cell lines
Human: HEK293T	ATCC	CRL-3216
Experimental models: Organisms/strains
S. mediterranea: asexual CIW4 clonal line	Lab stock	N/A
Oligonucleotides
primers	This paper	see Supplementary Table S1
Recombinant DNA
pGEM-T	Promega	Cat#A3600
pCDNA 3.1	ThermoFisher	Cat#V790-20
pCMV3flag8HOIL-1L	Fu et al. 121	Addgene Cat#50016
pGEX-XIAP	Yang et al. 120	Addgene Cat#8340
pGEM-Gypsy8	This Paper	N/A
pGEM-BURRO2	This Paper	N/A
pGEM-B1-5'MATH	This Paper	N/A
pGEM-SLF9	This Paper	N/A
pGEM-BURRO1-IAP	This Paper	N/A
pGEX-BIAP-GST	This Paper	N/A
pcDNA-IAP2-3xflag	This Paper	N/A
pcDNA-B-IAP-3xflag	This Paper	N/A
pcDNA-Caspase9038-3xflag	This Paper	N/A
pcDNA-BIAP	This Paper	N/A
pcDNA-Caspase9038-3xHA	This Paper	N/A
pcDNA-Caspase10708-3xHA	This Paper	N/A
pcDNA-Caspase11193-3xHA	This Paper	N/A
Software and algorithms
RepeatMasker (v4.1.2-p1-foss-2020b)	Smit et al.105	https://www.repeatmasker.org/
MAFFT v7	Katoh et al.106	https://mafft.cbrc.jp/alignment/software/source.html
MEGA11	Tamura et al.107	https://www.megasoftware.net
SeqKit (v2.8.1)	Shen et al.108	https://bioinf.shenwei.me/seqkit/
EMBOSS Needle	Rice et al.109	https://www.ebi.ac.uk/jdispatcher/psa/emboss_needle
BLAST+ version 2.15.0	Camacho et al.110	https://blast.ncbi.nlm.nih.gov/doc/blast-help/downloadblastdata.html#downloadblastdata
RAxML-NG v1.2.1	Kozlov et al.111	https://github.com/amkozlov/raxml-ng
MRBAYES version 3.2.7	Ronquist et al.112	https://nbisweden.github.io/MrBayes/download.html
interactive Tree of Life (iTOL)	Letunic et al.113	https://itol.embl.de/
VISTA	Frazer et al.115	https://genome.lbl.gov/vista/mvista/submit.shtml
fastp (version 0.21.0)	Chen et al.117	https://github.com/OpenGene/fastp
STAR (version 2.7.2a)	Dobin et al.118	https://github.com/alexdobin/STAR
FeatureCounts (version 1.6.4)	Liao et al.119	https://subread.sourceforge.net/
R Project for Statistical Computing v4.10	The R Foundation	https://www.r-project.org/
GraphPad Prism	GraphPad Software	https://www.graphpad.com/scientific-software/prism/
Planarian Image Quantification (PIQ) package v1.0.5	Allikka Parambil et al.54	https://gitlab.com/vanwolfswinkel/PIQ
Alphafold2	Jumper et al.103; Liu et al.102	https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb
ImageJ (FIJI)	Schindelin et al.96	https://fiji.sc

Highlights.

• Burro1 is a composite transposon that escaped inactivation for 100 million years

• Burro1 included an anti-apoptotic host gene (B-iap) that yields functional protein

• Burro1 (including B-iap) transcription is upregulated during stress

• Stress-induced B-IAP enhances stem cell survival and planarian regeneration