TIM29 is required for enhanced stem cell activity during regeneration in the flatworm Macrostomum lignano

TIM29 is a mitochondrial inner membrane protein that interacts with the protein import complex TIM22. TIM29 was shown to stabilize the TIM22 complex but its biological function remains largely unknown. Until recently, it was classified as one of the Domain of Unknown Function (DUF) genes, with a conserved protein domain DUF2366 of unclear function. Since characterizing DUF genes can provide novel biological insight, we used previously established transcriptional profiles of the germline and stem cells of the flatworm Macrostomum lignano to probe conserved DUFs for their potential role in germline biology, stem cell function, regeneration, and development. Here, we demonstrate that DUF2366/TIM29 knockdown in M. lignano has very limited effect during the normal homeostatic condition but prevents worms from adapting to a highly proliferative state required for regeneration.

www.nature.com/scientificreports/ human Translocase of the Inner Membrane 22 (TIM22) complex in HEK cells 13,14 . It was demonstrated that DUF2366 is required for maintaining the structural integrity and the assembly of the TIM22 complex, which mediates the import and insertion of hydrophobic proteins into the mitochondrial inner membrane 13,14 . Consequently, DUF2366 was renamed TIM29 13,14 . In addition, it was suggested that TIM29 contacts the Translocase of the Outer Membrane (TOM) complex, enabling transport of hydrophobic carrier substrates across the aqueous intermembrane space 13 . Interestingly, both papers studied the effect of TIM29 RNA interference (RNAi) on HEK cell proliferation, and reported contradicting results. While Kang et al. did not observe a significant effect of hTIM22 knockdown on cell proliferation 13 , Callegari et al. 14 observed a significantly decreased cell proliferation. In other words, the importance of TIM29 for cell proliferation remained unclear.
Here, we identify the DUFs conserved in M. lignano and demonstrate the crucial role of Mlig-DUF2366/ TIM29 for adapting to highly proliferative conditions during whole-body regeneration by means of RNA interference studies.

Identification of uncharacterized proteins in M. lignano.
To facilitate the discovery of novel genes involved in stem cell function, germline biology, regeneration, and development, we identified all genes in the M. lignano genome-guided transcriptome assembly Mlig_3_7_DV1_v3 12 encoding uncharacterized proteins (Suppl. Table 1). Due to partial genome duplication and redundancy, very closely related genes were grouped using Corset 15 into so-called transcript clusters for the downstream analysis 12,16 . Of the 820 identified DUF transcript clusters, 274 have identifiable homologs in human. Based on the expression level of the DUFs in different conditions and using previously established neoblast and germline transcriptional signatures 11 , categories were provided to predict their functional role in stem cells, the germline, regeneration, and development ( Table 1). The value of this candidate list was tested with a pilot RNA interference (RNAi) screen of three randomly chosen genes coding uncharacterized proteins: DUF2315 (Mlig002791.g5), UPF0197 (Mlig006314.g7), and DUF2366 (Mlig032364.g1). The screen focused on repeated tail regeneration of M. lignano (Fig. 1a), which depends on functional neoblasts. Knockdown of one of the three genes, Mlig-DUF2366, resulted in a reproducible phenotype. After 3 cycles of regeneration within 28 days, all (100%) DUF2366(RNAi) worms failed to regenerate new www.nature.com/scientificreports/ tissue, while all (100%) gfp(RNAi) worms, representing the negative control, successfully regenerated the tail (Fig. 1c). Interestingly, at least three repeated amputations of the tail-plate are necessary to induce this phenotype in 100% of the DUF2366(RNAi) worms. After a single tail-amputation, DUF2366(RNAi) worms were still able to regenerate a tail (Fig. 1b), and two tail-amputations demonstrated a variable degree of regeneration between DUF2366(RNAi) worms. Taken together, this suggests that without Mlig-DUF2366 expression, worms have limited regenerative abilities. Based on these results, we decided to further characterize DUF2366 in M. lignano, focusing on its requirement for stem cell function and regeneration.

M. lignano DUF2366 has an enriched expression in neoblasts and is homologous to TIM29. The
Mlig-DUF2366 gene has three nearly identical loci in the Mlig_3_7_DV1 genome assembly 12 : Mlig032364.g1, Mlig015320.g2, and Mlig018840.g2. These loci represent different gene copies emerged due to a recent wholegenome duplication and a duplication of the large chromosome of the M. lignano DV1 line 16,17 . All three Mlig-DUF2366 loci have strong homology to the TIM29 protein superfamily members (Pfam, PF10171) from diverse Metazoa (Suppl. Fig. 1), and will therefore be called Mlig-TIM29.
The previously obtained transcriptional profiles of sorted cells 11 demonstrated that Mlig-TIM29 transcripts have an elevated expression in proliferating somatic neoblasts compared to differentiated cells (Suppl. Table 1). Interestingly, according to the online PlanMine resource 18 , the Schmidtea mediterranea homolog, dd_Smed_ v6_9413_0_1 (Suppl. Fig. 1), also has a higher expression in X1 cells (cycling stem cells), compared to X2 (progenitors) and Xins cells (differentiated cells), and is included in the 'Stem cells versus differentiated cells_low stringency'-list (Suppl. Fig. 2). This suggests that elevated expression of TIM29 in neoblasts is conserved in multiple flatworm species. Online tools based on planarian single-cell sequencing [19][20][21] further confirm that TIM29 is predominantly expressed in clusters of neoblasts, including the cNeoblasts, and progenitors. Compared to the neoblasts/progenitors, differentiated cell-types have lower, but varying, expression levels (Suppl. Fig. 2).
The online MitoFates tool 22 predicts that the translated protein sequence of the Mlig-TIM29 transcripts contain a mitochondrial presequence and TOM20 recognition motifs, indicating a mitochondrial localization of the protein. MitoFates also predicts a mitochondrial presequence for dd_Smed_v6_9413_0_1 (PlanMine) and human TIM29 (GeneBank accession: NP_612367.1) translated protein sequences (data not shown).

Experimental setup to study the role of Mlig-TIM29 in neoblasts and regeneration. To inves-
tigate the potential role of Mlig-TIM29 in stem cell function and regeneration, a set of RNAi experiments was performed using M. lignano. In total, four specific experimental 'classes' were characterized: gfp(RNAi) uncut, gfp(RNAi) cut, Mlig-TIM29(RNAi) uncut, and Mlig-TIM29(RNAi) cut (Fig. 2). As GFP is not expressed in wild type worms, the gfp(RNAi) classes represent the negative control. The uncut conditions represent worms in which proliferation is only required for cell turnover during adult tissue homeostasis, and the production of gametes. In the cut conditions, the body was amputated by cutting worms between the pharynx and testes after 1 week of RNAi treatment, inducing regeneration of the whole body. Flatworm regeneration is a convenient readout for stem cell functionality, as it requires neoblast proliferation, migration, and differentiation.
Different measurements were performed on all four RNAi classes (Fig. 2). First, the morphology of all four classes, and the regeneration capacity of cut worms were assessed. Second, the number of mitotic cells was quantified. Third, gene expression was studied by means of RNA sequencing (Suppl. Table 2). Both mitotic labeling and RNA-seq were performed on the tenth day of RNAi, which represents the second day of regeneration in the cut worms. Fourth, to interpret the changes in gene expression, Gene Ontology (GO) Term analysis of differentially expressed genes was performed.
Mlig-TIM29 knockdown during homeostasis does not result in a prominent phenotype. After 4 weeks of RNAi, the majority of uncut Mlig-TIM29(RNAi) worms (> 95%) still had a similar morphology as the uncut gfp(RNAi) worms (Fig. 3a). A limited number of individuals (< 5%), however, showed an impaired maintenance of the body by e.g. degeneration of the gonads, degeneration of the rostrum, a shrinking size, and appearance of small bulges (Fig. 3a), which often resulted in death of the individual. At the tenth day of RNAi, the number of mitotic cells, representing dividing somatic neoblasts and germline cells, was not significantly different between homeostatic control and Mlig-TIM29(RNAi) worms (p = 0.610, ANOVA and post-hoc Tukey test) (Fig. 4a, Suppl. Fig. 3). This demonstrates that Mlig-TIM29(RNAi) worms can maintain the proliferation rate required for cellular turnover during homeostasis. Differential gene expression analysis indicated that uncut Mlig-TIM29(RNAi) and uncut gfp(RNAi) worms only show minimal differences at the molecular level, as only 23 significantly differentially expressed genes were identified (Fig. 4b). Importantly, the two most downregulated genes, for which the expression decreased more than 10 times (log 2 = − 3.5), correspond to the Mlig-TIM29 transcripts. This confirms the high efficiency of the RNAi knockdown of Mlig-TIM29. In conclusion, efficient knockdown of Mlig-TIM29 in uncut worms did not result in a consistent prominent phenotype within the timeframe of 4 weeks.
Mlig-TIM29 knockdown impairs whole-body regeneration. The regenerative ability was assessed at 1 and 2 weeks after amputation of the body, and was clearly inhibited by knockdown of Mlig-TIM29 at both time points. One week after amputation, all Mlig-TIM29(RNAi) worms (100%) had a smaller tail than the negative controls, although a large variation could be observed between different individuals. In worst case, there was a complete lack of regeneration, while in best case worms regenerated a small tail plate (Fig. 3b). With advancing time, the phenotype became more apparent. Two weeks after amputation, the gfp(RNAi) worms were completely regenerated and resembled young adults. In contrast, all Mlig-TIM29(RNAi) worms (100%) were small and disproportionate. Due to impaired regenerative tissue remodeling, the location of amputation could still be www.nature.com/scientificreports/ observed. The morphology of the tail still varied from lacking to a small tail plate. In several cases, the appearance of bulges could be observed (Fig. 3b).
Mlig-TIM29 is required for adapting to regenerative conditions. The observation that knockdown of Mlig-TIM29 results in a consistent phenotype only during triple tail-regeneration and whole-body regeneration indicates that worms cannot adapt to highly proliferative conditions during RNAi treatment. This is confirmed by differential gene expression analysis, and the quantification of mitotic neoblasts.
In gfp(RNAi) worms, genes with an upregulated expression due to regeneration are enriched for neoblast transcripts (1.37-fold; p < 10 −12 , Pearson's Chi-squared test with Yates' continuity correction). In contrast, in Mlig-TIM29(RNAi) worms, neoblast transcripts are depleted among genes with increased expression due to www.nature.com/scientificreports/ regeneration (0.44-fold, p < 10 −12 ) (Suppl. Table 3). This difference indicates that Mlig-TIM29 is required for the neoblast response during whole-body regeneration. This observation is confirmed by quantification of mitotic neoblasts in the amputated head fragments and in the corresponding region of the body, anterior of the testes, in uncut worms. In gfp(RNAi) worms, regeneration causes a significant increase in the number of mitotic neoblasts, demonstrating increased proliferation during regeneration (p = 0.001, ANOVA and post-hoc Tukey test) (Fig. 4a). This mitotic activity is mainly located in the blastema-region (Suppl. Fig. 3). During Mlig-TIM29 knockdown, inducing regeneration does not significantly increase the number of mitotic neoblasts (p = 0.999, ANOVA and post-hoc Tukey test) (Fig. 4a, Suppl. Fig. 3).
To replace missing body structures, cells in the blastema have to differentiate 23 . Both in gfp(RNAi) worms and Mlig-TIM29(RNAi) worms, genes with an upregulated expression during regeneration are enriched for GO terms related to differentiation (Suppl. Table 4). This suggests that the regenerative phenotype due to Mlig-TIM29 knockdown is not caused by inhibited differentiation.
An important, but less understood aspect of regeneration involves anatomical remodeling to restore scale and proportion, and to allow the integration of new and old tissues 23,24 . Apoptosis has been shown to be a central mechanism for this 24 . In gfp(RNAi) worms, genes with an increased expression during regeneration are enriched for GO terms related to intrinsic and extrinsic apoptotic signaling pathways (Suppl. Table 4). In Mlig-TIM29(RNAi) worms, however, genes with a differential expression during regeneration are not enriched for GO terms related to cell death (Suppl. Table 4). These results suggest that Mlig-TIM29 knockdown has an effect on apoptosis during regeneration, which could explain the limited remodeling. In conclusion, by impacting apoptosis and neoblast-proliferation, Mlig-TIM29 knockdown limits the regenerative capacity.
Knockdown of Mlig-TIM29 has a global cellular impact in regenerating worms. To better understand the impaired regenerative response during Mlig-TIM29(RNAi), we compared gene expression between cut Mlig-DUF2366(RNAi) and cut gfp(RNAi) worms in more detail. In total, 3854 significantly differentially expressed genes were identified, of which 1827 were upregulated and 2027 were downregulated (Fig. 4b). GO term analysis shows that, during whole-body regeneration, multiple molecular biological processes are affected due to knockdown of Mlig-TIM29. Interestingly, however, biological clusters of significantly enriched GO terms are only found for downregulated genes. Clusters of GO terms related to cell division further confirm that Mlig- www.nature.com/scientificreports/ TIM29(RNAi) worms are not able to obtain high levels of proliferation to regenerate the body. In addition, large clusters of GO terms related to translation and protein transport, metabolism, transcription regulation and RNA processing, and mitochondrial function are found (Fig. 5). Thus, knockdown of Mlig-TIM29 has a global effect, impacting multiple basic molecular processes and organelles in regenerating worms.

Discussion
All data together demonstrates that knockdown of one gene, Mlig-TIM29, is enough to impair whole-body regeneration. Computational analysis indicated a mitochondrial localization of the TIM29 protein in M. lignano, which is consistent with the recent renaming of the human DUF2366 as TIM29 based on its characterization as an inner mitochondrial membrane protein in HEK cells 13,14 . This mitochondrial localization and function correlates with the enrichment of several GO term processes and components related to mitochondria in the differentially expressed genes between cut Mlig-TIM29(RNAi) and cut gfp(RNAi) worms (Fig. 5). The observed global changes in metabolism, translation, transcription regulation, DNA repair, stimuli response, and mitochondrial function (Fig. 5) suggest that knockdown of Mlig-TIM29 induces a state of cellular stress which can inhibit the required molecular response for successful regeneration and growth. As a result, worms are not able to increase the proliferation rate required for whole-body regeneration, which is shown at both the cellular and molecular level (Fig. 4). In addition, GO term analysis suggests that Mlig-TIM29 knockdown has an effect on apoptosis during regeneration, explaining the limited tissue-remodeling (Fig. 3b, Suppl. Table 4). Flatworm studies of regenerative cell death are very limited [24][25][26] , but it has been shown in S. mediterranea that regenerative apoptosis occurs predominantly in differentiated cells, and does not depend on neoblasts and their proliferation 24 .
The opposite scenario, cell death triggering neoblast proliferation, has not been tested yet in flatworms 24,26 . The lack of tools makes it complicated to perform cellular studies of cell death and its crosstalk with proliferation during Mlig-TIM29(RNAi) and regeneration of M. lignano in general. Future development of transgenic tools to study diverse aspects of cells death in M. lignano will aid research of this important aspect of flatworm biology. Our findings fit with the increasing recognition of mitochondrial signaling as a key component to mediate stem cell function. The emerging picture is that mitochondria continuously integrate cellular and environmental cues to influence stem cell fate and activity, which enables organisms to adapt to the environmental changes [27][28][29] . Many questions remain, however, as the majority of published mitochondrial research focused on post-mitotic tissues, and the role of mitochondria in the context of stem cells has been largely neglected until recent. www.nature.com/scientificreports/ The development of a Mlig-TIM29(RNAi)-phenotype only during large scale regeneration fits with the description of DUF proteins having a function of which the importance can be limited to, or only observed during, specific conditions 1,3 . To unravel the function of DUFs, it is therefore important to study different conditions. M. lignano can be an appropriate model for this, as it provides an in vivo system enabling to study stem cells in their natural environment. Different conditions besides cellular turnover during adult homeostasis can be easily induced. Examples are different levels of regeneration by amputating different portions of the body, development, starvation, growth and even degrowth based on the available amount of food. Moreover, the expanding molecular toolbox and especially the recently developed methods of transgenesis will further facilitate in vivo studies to identify and characterize the function of Uncharacterized Proteins 16,30 . The here presented list of Uncharacterized Proteins presents an ideal starting point for selecting candidates. The value of this list is demonstrated as screening three candidates was sufficient for identifying a candidate which is required for adapting to highly proliferative conditions during regeneration. Homology detection and sequence analysis. To find a homology to other known protein families, nucleotide sequences of the transcribed M. lignano DUF2366 loci as well as S. mediterranea dd_Smed_ v6_9413_0_1 transcript sequence (PlanMine) were directly submitted to the NCBI Conserved Domain Search server 33 . Open reading frames (ORFs) analysis, multiple sequence alignments construction and visualization were done in Uinpro UGENE v34.0 34 . Amino acid sequences of TIM29 conserved domain family of other Metazoa species were obtained directly from Pfam (https ://pfam.xfam.org/famil y/PF101 71). Prediction of mitochondrial processing presequence and TOM20 recognition motifs was performed using the online MitoFates tool 22 submitting translated ORFs sequences.

Materials and methods
RNA interference. The production of dsRNA was performed following a previously published protocol 11,12 , and the primers of candidate genes are presented in (Suppl. Table 5). Candidate genes were knocked down by means of RNAi with double-stranded RNA delivered by soaking as previously described 11,35 . The RNAi soaking experiments were performed in 24-well plates in which diatoms were grown. Individual wells contained 300 µl of dsRNA solution (10 ng/µl in f/2 medium) in which 15 individuals were maintained. The preliminary RNAi-screen, lasted for 4 weeks, and the tail of worms was amputated after 1, 2, and 3 weeks of RNAi treatment. For the Mlig-TIM29(RNAi)-screen, worms of the regenerative condition were treated for 3 weeks, and the body was amputated by cutting the worms in the region between the testes and pharynx after 1 week of RNAi treatment. Worms of the homeostasis condition were treated for 4 weeks. In all RNAi treatments, animals were weekly transferred to fresh 24-well plates to ensure sufficient amounts of food. As a negative control, gfp dsRNA was used. To quantify the occurrence of phenotypes, a stereomicroscope was used to observe worms, and 3 independent RNAi experiments starting with 15 worms (total n = 45) were performed for each condition: gfp(RNAi) uncut, gfp(RNAi) cut, Mlig-TIM29(RNAi) uncut, and Mlig-TIM29(RNAi) cut. To illustrate the observed changes in morphology and regeneration capacity, photos were taken using an EVOS XL Core Imaging System (ThermoFisher). For this, worms were temporary relaxed in 1:1 f/2:MgCl 2 .6H 2 O (7.14%) in a small drop in a Petri dish.
Mitotic labeling. For both Mlig-TIM29(RNAi) and gfp(RNAi), cut and uncut worms were collected at the tenth day of RNAi treatment. In the cut condition, this time point represents 2 days after amputation. Mitotic labeling was performed as described before 6,11 . In short, worms were washed in f/2 medium, relaxed in 1:1 f/2:MgCl 2 ·6H 2 O (7.14%), and fixed in 4% paraformaldehyde (PFA) for 1 h at room temperature (RT). Afterwards, they were washed with PBS-T (PBS and 0.1% Triton X-100) and blocked with BSA-T (1% bovine serum albumin in PBS-T) for 30 min at RT. The primary anti-phospho histone H3 Antibody (Millipore) was diluted 1:250 in BSA-T and applied overnight at 4 °C, followed by washing with PBS-T at RT. Worms were then incubated in secondary goat anti-rabbit IgG Antibody conjugated with FITC (Millipore) which is diluted 1:150 in BSA-T, for 1 h at RT. After being washed with PBS-T, slides were mounted using Vectashield (Vector Laboratories US, Burlingame, CA). Mitotic cells were visualized using a Leica TCS SP8 confocal microscope and were counted through the entire Z-stack of the animals using the Cell counter plugin in ImageJ. For each of the four conditions, the number of mitotic cells was quantified for a total of 12 individuals obtained from two independent labeling-experiments (n = 2 * 6). To determine if the number of cells was significantly different between conditions, an ANOVA and post-hoc Tukey test were performed in SPSS.
Preparation and sequencing of RNA-seq libraries. Worms of the four different conditions (Mlig-TIM29(RNAi) uncut; Mlig-TIM29(RNAi) cut; gfp(RNAi) uncut; gfp(RNAi) cut) were collected at the tenth day of RNAi treatment. In the cut condition, this time point represents 2 days after amputation. For each condition, four replicates of 45 individuals each were rinsed with f/2 medium, suspended in 500 µl TRIzol reagent (Ambion) and stored at -80 °C.
RNA was extracted from the samples with the Direct-zol RNA MiniPrep Kit (Zymo Research), following the manufacturer's protocol. RNA-Seq libraries were made using the CEL-Seq2 protocol 36,37 , and as a first step a mix of RNA, primer, spike-in, and dNTPs was made. While this method was originally designed for single www.nature.com/scientificreports/ cells, it also works well with larger amounts of RNA. Sequencing was performed using the T-fill protocol 38 on an Illumina HiSeq 2500 machine.
Differential expression analysis of RNA-Seq data. Illumina reads were mapped to the M. lignano genome assembly Mlig_3_7 16 using STAR software v. 2.6.0c 39 and transcriptome annotation version Mlig_ RNA_3_7_v3 12 . The transcriptome quantification option of STAR was used to derive initial transcript counts, which were consolidated into transcript cluster counts using Corset 15 . Differential gene expression analysis was performed using generalized linear models implemented in edgeR software package 40 . Only transcript clusters that had at least 1 count per million in at least 3 samples were included in the analysis. FDR cutoff of 0.05 was used to establish statistically significant differentially expressed genes.

GO Term analysis.
For genes differentially expressed between various experimental conditions human homolog gene annotations were extracted from the previously annotated trascriptome Mlig_RNA_3_7_v3 12 .
The resulted list of M. lignano human homologs was then processed and analyzed using a custom script written in R programming language. Libraries "org.Hs.eg.db" 41 and "clusterProfiler" 42 were used for the GO term analysis, applying the function enrichGO with the following parameters: [List of Human ENTREZ gene IDs], OrgDb = org.Hs.eg.db, ont = "BP", pvalueCutoff = 0.01, qvalueCutoff = 0.01, pAdjustMethod = "BH". The results of the analysis were visualized as a graph using emapplot function, showing all categories and coloring nodes by their adjusted p-values. The graph was exported to PDF and manually processed in Inkscape v.0.92 vector graphics software (https ://inksc ape.org/) to apply additional annotations.