Regeneration-capable flatworms are informative research models to study the mechanisms of stem cell regulation, regeneration, and tissue patterning. However, the lack of transgenesis methods considerably hampers their wider use. Here we report development of a transgenesis method for Macrostomum lignano, a basal flatworm with excellent regeneration capacity. We demonstrate that microinjection of DNA constructs into fertilized one-cell stage eggs, followed by a low dose of irradiation, frequently results in random integration of the transgene in the genome and its stable transmission through the germline. To facilitate selection of promoter regions for transgenic reporters, we assembled and annotated the M. lignano genome, including genome-wide mapping of transcription start regions, and show its utility by generating multiple stable transgenic lines expressing fluorescent proteins under several tissue-specific promoters. The reported transgenesis method and annotated genome sequence will permit sophisticated genetic studies on stem cells and regeneration using M. lignano as a model organism.
Animals that can regenerate missing body parts hold clues to advancing regenerative medicine and are attracting increased attention1. Significant biological insights on stem cell biology and body patterning were obtained using free-living regeneration-capable flatworms (Platyhelminthes) as models2–4. The most often studied representatives are the planarian species Schmidtea mediterranea2 and Dugesia japonica5. Many important molecular biology techniques and resources are established in planarians, including fluorescence-activated cell sorting, gene knockdown by RNA interference, in situ hybridization, and genome and transcriptome assemblies4. One essential technique still lacking in planarians; however, is transgenesis, which is required for in-depth studies involving e.g., gene overexpression, dissection of gene regulatory elements, real-time imaging and lineage tracing. The reproductive properties of planarians, including asexual reproduction by fission and hard non-transparent cocoons containing multiple eggs in sexual strains, make development of transgenesis technically challenging in these animals.
More recently, a basal flatworm Macrostomum lignano (Macrostomorpha) emerged as a model organism that is complementary to planarians6–9. The reproduction of M. lignano, a free-living marine flatworm, differs from planarians, as it reproduces by laying individual fertilized one-cell stage eggs. One animal lays ~1 egg per day when kept in standard laboratory conditions at 20 °C. The eggs are around 100 microns in diameter, and follow the archoophoran mode of development, having yolk-rich oocytes instead of supplying the yolk to a small oocyte via yolk cells10. The laid eggs have relatively hard shells and can easily be separated from each other with the use of a fine plastic picker. These features make M. lignano eggs easily amenable to various manipulations, including microinjection11. In addition, M. lignano has several convenient characteristics, such as ease of culture, transparency, small size, and a short generation time of three weeks6,7. It can regenerate all tissues posterior to the pharynx, and the rostrum12. This regeneration ability is driven by stem cells, which in flatworms are called neoblasts3,4,13. Recent research in planarians has shown that the neoblast population is heterogeneous and consists of progenitors and stem cells14,15. The true pluripotent stem cell population is, however, not identified yet.
Here we present a method for transgenesis in M. lignano using microinjection of DNA into single-cell stage embryos and demonstrate its robustness by generating multiple transgenic tissue-specific reporter lines. We also present a significantly improved genome assembly of the M. lignano DV1 line and an accompanying transcriptome assembly and genome annotation. The developed transgenesis method, combined with the generated genomic resources, will enable new research avenues on stem cells and regeneration using M. lignano as a model organism, including in-depth studies of gene overexpression, dissection of gene regulatory elements, real-time imaging and lineage tracing.
Microinjection and random integration of transgenes
M. lignano is an obligatorily non-self-fertilizing simultaneous hermaphrodite (Fig. 1a) that produces substantial amounts of eggs (Fig. 1b, c). We reasoned that microinjection approaches used in other model organisms, such as Drosophila, zebrafish and mouse, should also work in M. lignano eggs (Fig. 1d, Supplementary Movie 1). First, we tested how the egg handling and microinjection procedure itself impacts survival of the embryos (Supplementary Table 1). Separating the eggs laid in clumps and transferring them into new dishes resulted in a 17% drop in hatching rate, and microinjection of water decreased survival by a further 10%. Thus, in our hands >70% of the eggs can survive the microinjection procedure (Supplementary Table 1). When we injected fluorescent Alexa 555 dye, which can be used to track the injected material, about 50% of the eggs survived (Supplementary Table 1). For this reason, we avoided tracking dyes in subsequent experiments. Next, we injected in vitro synthesized mRNA encoding green fluorescent protein (GFP) and observed its expression in all successfully injected embryos (n > 100) within 3 h after injection (Fig. 1e), with little to no autofluorescence detected in either embryos or adult animals (Supplementary Fig. 1). The microinjection technique can thus be used to deliver biologically relevant materials into single-cell stage eggs with a manageable impact on the survival of the embryos.
Macrostomum lignano embryos are amenable to microinjection. a Schematic morphology and a bright-field image of an adult M. lignano animal. b Clump of fertilized eggs. c DIC image of a one-cell stage embryo. d Microinjection into a one-cell stage embryo....
To investigate whether exogenous DNA constructs can be introduced and expressed in M. lignano, we cloned a 1.3 kb promoter region of the translation elongation factor 1 alpha (EFA) gene and made a transcriptional GFP fusion in the Minos transposon system (Supplementary Fig. 2a). Microinjection of the Minos::pEFA::eGFP plasmid with or without Minos transposase mRNA resulted in detectable expression of GFP in 5–10% of the injected embryos (Supplementary Fig. 2c). However, in most cases GFP expression was gradually lost as the animals grew (Supplementary Fig. 2f), and only a few individuals transmitted the transgene to the next generation. From these experiments we established the HUB1 transgenic line with ubiquitous GFP expression, which recapitulates expression of the EFA gene determined by in situ hybridization (Supplementary Fig. 2d, e). Stable transgene transmission in the HUB1 line has been observed for over 50 generations16,17.
The expected result for transposon-mediated transgenesis is genomic integration of the fragment flanked by transposon inverted terminal repeats. However, plasmid sequences outside the terminal repeats, including the ampicillin resistance gene, were detected in the HUB1 line, suggesting that the integration was not mediated by Minos transposase. Furthermore, southern blot analysis revealed that HUB1 contains multiple transgene copies (Supplementary Fig. 2g). We next tried a different transgenesis strategy using meganuclease I-SceI18 to improve transgenesis efficiency (Supplementary Fig. 2b). We observed a similar 3–10% frequency of initial transgene expression, and only two instances of germline transmission, one of which resulted from the negative control experiment without co-injected meganuclease protein (Supplementary Fig. 2c). These results suggest that I-SceI meganuclease does not increase efficiency of transgenesis in M. lignano, but instead that exogenous DNA can be integrated in the genome by non-homologous recombination using the endogenous DNA repair machinery.
Improvement of integration efficiency
The frequency of germline transgene transmission in the initial experiments was <0.5% of the injected eggs, while transient transgene expression was observed in up to 10% of the cases (Supplementary Fig. 2c, f). We hypothesized that mosaic integration or mechanisms similar to extrachromosomal array formation in C. elegans19 might be at play in cases of transient gene expression in M. lignano. We next tested two approaches used in C. elegans to increase the efficiency of transgenesis: removal of vector backbone and injection of linear DNA fragments20, and transgene integration by irradiation19. Injection of PCR-amplified vector-free transgenes resulted in the germline transmission in 5 cases out of 269 injected eggs, or 1.86% (Table 1), and the stable transgenic line NL1 was obtained during these experiments (Fig. 2a). In this line, the GFP coding sequence was optimized for M. lignano codon usage. While we did not observe obvious differences in expression levels between codon-optimized and non-optimized GFP sequences, we decided to use codon-optimized versions in all subsequent experiments.
Efficiency of transgenesis with different reporter constructs and treatments
Ubiquitously expressed elongation factor 1 alpha promoter transgenic lines. a NL1 line expressing enchanced GFP (eGFP). b NL3 line expressing codon-optimized Cherry (oCherry). c NL20 line expressing codon-optimized nuclear localized H2B::oGFP fusion....
M. lignano is remarkably resistant to ionizing radiation, and a dose as high as 210 Gy is required to eliminate all stem cells in an adult animal8,21. We reasoned that irradiation of embryos immediately after transgene injection might stimulate non-homologous recombination and increase integration rates. Irradiation dose titration revealed that M. lignano embryos are less resistant to radiation than adults and that a 10 Gy dose results in hatching of only 10% of the eggs, whereas >90% of eggs survive a still substantial dose of 2.5 Gy (Supplementary Table 2). Irradiating injected embryos with 2.5 Gy resulted in 1–8% germline transmission rate for various EFA promoter constructs in both plasmid and vector-free forms (Table 1). The stable transgenic line NL3 expressing codon-optimized red fluorescent protein Cherry was obtained in this way (Fig. 2b), demonstrating that ubiquitous expression of fluorescent proteins other than GFP is also possible in M. lignano. Finally, to test nuclear localization of the reporter protein, we fused GFP with a partial coding sequence of the histone 2B (H2B) gene as described previously22. The injection of the transgene fragment followed by irradiation demonstrated 5% transgenesis efficiency (Table 1), and the stable NL20 transgenic line with nuclear GFP localization was established (Fig. 2c).
Genome assembly and annotation
To extend the developed transgenesis approach to promoters of other genes, an annotated genome assembly of M. lignano was required. Toward this, we have generated and sequenced 29 paired-end and mate-pair genomic libraries of the DV1 line using 454 and Illumina technologies (Supplementary Table 3). Assembling these data using the MaSuRCA genome assembler23 resulted in a 795 Mb assembly with N50 scaffold size of 11.9 kb. While this assembly was useful for selecting several novel promoter regions, it suffered from fragmentation. In a parallel effort, a PacBio-based assembly of the DV1 line, termed ML2, was recently published9. The ML2 assembly is 1040 Mb large and has N50 contig size of 36.7 kb and NG50 contig size of 64.5 kb when adjusted to the 700 Mb genome size estimated from k-mer frequencies9. We performed fluorescence-based genome size measurements and estimated that the haploid genome size of the DV1 line is 742 Mb (Supplementary Fig. 3d,e,f). It was recently demonstrated that M. lignano can have a polymorphic karyotype, where in addition to the basal 2n = 8 karyotype, also animals with aneuploidy for the large chromosome, with 2n = 9 and 2n = 10 exist24. We confirmed that our laboratory culture of the DV1 line has predominantly 2n = 10 and 2n = 9 karyotypes (Supplementary Fig. 3a, b) and estimated that the size of the large chromosome is 240 Mb (Supplementary Fig. 3f). In contrast, an independently established M. lignano wild-type line NL10 has the basal karyotype 2n = 8 and does not show detectable variation in chromosome number (Supplementary Fig. 3c,d). This line, however, was established only recently and was not a part of the genome sequencing effort.
We re-assembled the DV1 genome from the generated Illumina and 454 data and the published PacBio data9 using the Canu assembler25 and SSPACE scaffolder26. The resulting Mlig_3_7 assembly is 764 Mb large with N50 contig and scaffold sizes of 215.2 Kb and 245.9 Kb, respectively (Table 2), which is greater than threefold continuity improvement over the ML2 assembly. To compare the quality of the ML2 and Mlig_3_7 assemblies, we used the genome assembly evaluation tool REAPR, which identifies assembly errors without the need for a reference genome27. According to the REAPR analysis, the Mlig_3_7 assembly has 63.95% of error-free bases compared to 31.92% for the ML2 assembly and 872 fragment coverage distribution (FCD) errors within contigs compared to 1871 in the ML2 assembly (Supplementary Fig. 4a). Another genome assembly evaluation tool, FRCbam, which calculates feature response curves for several assembly parameters28, also shows better overall quality of the Mlig_3_7 assembly (Supplementary Fig. 4b). Finally, 96.9% of transcripts from the de novo transcriptome assembly MLRNA1509048 can be mapped on Mlig_3_7 (>80% identity, >95% transcript length coverage), compared to 94.88% of transcripts mapped on the ML2 genome assembly, and among the mapped transcripts more have intact open reading frames in the Mlig_3_7 assembly than in ML2 (Supplementary Fig. 4c). Based on these comparisons, the Mlig_3_7 genome assembly represents a substantial improvement in both continuity and base accuracy over the ML2 assembly.
Characteristics of Mlig_3_7 genome assembly
More than half of the genome is repetitive, with LTR retrotransposons and simple and tandem repeats accounting for 21 and 15% of the genome, respectively (Supplementary Table 4). As expected from the karyotype of the DV1 line, which has additional large chromosomes, the Mlig_3_7 assembly has substantial redundancy, with 180 Mb in duplicated non-repetitive blocks that are longer than 500 bp and at least 95% identical. When repeat-annotated regions are included in the analysis, the duplicated fraction of the genome rises to 312 Mb.
Since genome-guided transcriptome assemblies are generally more accurate than de novo transcriptome assemblies, we generated a new transcriptome assembly based on the Mlig_3_7 genome assembly using a combination of the StringTie29 and TACO30 transcriptome assemblers, a newly developed TBONE gene boundary annotation pipeline, previously published RNA-seq datasets8,31 and the de novo transcriptome assembly MLRNA1509048. Since many M. lignano transcripts are trans-spliced8,9, we extracted reads containing trans-splicer leader sequences from raw RNA-seq data and mapped them to the Mlig_3_7 genome assembly after trimming the trans-splicing parts. This revealed that many more transcripts in M. lignano are trans-spliced than was previously appreciated from de novo transcriptome assemblies (6167 transcripts in Grudniewska et al.8, 7500 transcripts in Wasik et al.9, 28,273 in this study, Table 3). We also found that almost 7% of the assembled transcripts are in fact precursor mRNAs, i.e., they have several trans-splicing sites and encode two or more proteins (Table 3, Supplementary Fig. 5a). Therefore, in the transcriptome assembly we distinguish between transcriptional units and genes transcribed within these transcriptional units. For this, we developed computational pipeline TBONE (Transcript Boundaries based ON experimental Evidence), which relies on experimental data, such as trans-splicing and polyadenylation signals derived from RNA-seq data, to ‘cut’ transcriptional units and establish boundaries of mature mRNAs (Supplementary Fig. 5a). The new genome-guided transcriptome assembly, Mlig_RNA_3_7_DV1.v1, has 66,777 transcriptional units, including duplicated copies and alternative forms, which can be collapsed to 33,715 non-redundant transcripts when clustered by 95% global sequence identity (Table 3). These transcriptional units transcribe 72,846 genes, of which 44,328 are non-redundant, 38.8% are trans-spliced and 79.98% have an experimentally defined poly(A) site (Table 3). The non-redundant transcriptome has TransRate scores of 0.4360 and 0.4797 for transcriptional units and gene sequences, respectively, positioning it among the highest quality transcriptome assemblies32. The transcriptome is 98.1% complete according to the Benchmarking Universal Single-Copy Orthologs33, with only 3 missing and 3 fragmented genes (Table 3).
Characteristics of Mlig_RNA_3_7_DV.v1 transcriptome assembly
The Mlig_RNA_3_7_DV1 transcriptome assembly, which incorporates experimental evidence for gene boundaries, greatly facilitates selection of promoter regions for transgenesis. Furthermore, we previously generated 5′-enriched RNA-seq libraries from mixed stage populations of animals8 using RAMPAGE34. In our hands, the RAMPAGE signal is not sufficiently localized around transcription start sites to be used directly by the TBONE pipeline, but it can be very useful for determining transcription starts during manual selection of promoter regions for transgenesis (Supplementary Fig. 5b, c). We used the UCSC genome browser software35 to visualize genome structure and facilitate design of new constructs for transgenesis (Supplementary Fig. 5). The M. lignano genome browser, which integrates genome assembly, annotation and RNA-seq data, is publicly accessible at http://gb.macgenome.org.
Tissue-specific transgenic lines
Equipped with the annotated M. lignano genome and the developed transgenesis approach, we next set to establish transgenic lines expressing tissue-specific reporters. For this, we selected homologs of the MYH6, APOB, ELAV4, and CABP7 genes, for which tissue specificity in other model organisms is known and upstream promoter regions can be recognized based on genome annotation and gene boundaries (Supplementary Fig. 5). Similar to the EFA promoter, in all cases the transgenesis efficiency was in the range of 1–5% of the injected eggs (Table 1) and stable transgenic lines were obtained (Fig. 3). Expression patterns were as expected from prior knowledge and corroborated by the whole mount in situ hybridization results: the MYH6::GFP is expressed in muscle cells, including muscles within the stylet (Fig. 3a, Supplementary Movie 2); APOB::GFP is gut-specific (Fig. 3b); ELAV4::GFP is testis-specific, including the sperm, which is accumulated in the seminal vesicle (Fig. 3c); and CABP7::GFP is ovary-specific and is also expressed in developing eggs (Fig. 3d). Finally, we made a double-reporter construct containing ELAV4::oNeonGreen and CABP7::oScarlet-I in a single plasmid (Fig. 3e). mNeonGreen36 and mScarlet37 are monomeric yellow–green and red fluorescent proteins, respectively, with the highest reported brightness among existing fluorescent proteins. The transgenesis efficiency with the double-reporter construct was comparable to other experiments (Table 1), and transgenic line NL24 expressing codon-optimized mNeonGreen (oNeonGreen) in testes and codon-optimized mScarlet-I (oScarlet) in ovaries was established (Fig. 3e), demonstrating the feasibility of multi-color reporters in M. lignano. The successful generation of stable transgenic reporter lines for multiple tissue-specific promoters validates the robustness of the developed transgenesis method and demonstrates the value of the generated genomic resource.
Tissue-specific promoter transgenic lines. a NL9 line expressing GFP under the muscle-specific promoter of the MYH6 gene. Zoom in—detailed images of the body wall (top) and stylet (bottom); In situ—whole-mount in situ hybridization expression...
Identification of transgene integration sites
To directly demonstrate that transgenes integrate into the M. lignano genome and to establish genomic locations of the integration sites, we initially attempted to identify genomic junctions by inverse PCR with outward-oriented transgene-specific primers (Supplementary Fig. 6a) in the NL7 and NL21 transgenic lines. However, we found that in both cases short products of ~200 nt are preferentially and specifically amplified from genomic DNA of the transgenic lines (Supplementary Fig. 6b, c). The size of the PCR products can be explained by formation of tandem transgenes (Supplementary Fig. 6a), and sequencing confirmed that this is indeed the case (Supplementary Fig. 6d). Next, we used the Genome Walker approach, in which genomic DNA is digested with a set of restriction enzymes, specific adapters are ligated and regions of interest are amplified with transgene-specific and adapter-specific primers. Similarly, many of the resulting PCR products turned out to be transgene tandems. But in the case of the NL21 line we managed to establish the integration site on one side of the transgene (Supplementary Fig. 6e), namely at position 45,440 in scaf3369 (Mlig_3_7 assembly) in the body of a 2-kb long LTR retrotransposon, 10.5 kb downstream from the end of the Mlig003479.g3 gene and 2.5 kb upstream from the start of the Mlig028829.g3 gene.
Transgene expression in regenerating animals
Our main rationale for developing M. lignano as a new model organism is based on its experimental potential to study the biology of regenerative processes in vivo in a genetically tractable organism. Therefore, it is essential to know whether regeneration could affect transgene stability and behavior. Toward this, we monitored transgene expression during regeneration in the testis- and ovary-specific transgenic lines NL21 and NL23, respectively (Fig. 4). Adult animals were amputated anterior of the gonads and monitored for 10 days. In both transgenic lines regeneration proceeded normally and no GFP expression was observed in the first days of regeneration (Fig. 4). Expression in ovaries was first detected at day 8 after amputation, and in testes at day 10 after amputation (Fig. 4). Thus, tissue-specific transgene expression is restored during regeneration, as expected for a regular genomic locus.
Transgene expression during regeneration. a Testes-specific transgenic line NL23. b Ovaries-specific transgenic line NL22. BF—bright-field, FITC—FITC channel. Day 0—animals immediately after amputation, both head and tail regions...
Free-living regeneration-capable flatworms are powerful model organisms to study mechanisms of regeneration and stem cell regulation2,4. Currently, the most popular flatworms among researchers are the planarian species S. mediterranea and D. japonica4. A method for generating transgenic animals in the planarian Girardia tigrina was reported in 200338, but despite substantial ongoing efforts by the planarian research community it has thus far not been reproduced in either S. mediterranea or D. japonica. The lack of transgenesis represents a significant experimental limitation of the planarian model systems. Primarily for this reason we focused on developing an alternative, non-planarian flatworm model, Macrostomum lignano. We reasoned that the fertilized one-cell stage eggs, which are readily available in this species, will facilitate development of the transgenesis method, leveraging the accumulated experience on transgenesis in other model organisms.
In this study, we demonstrate a reproducible transgenesis approach in M. lignano by microinjection and random integration of DNA constructs. Microinjection is the method of choice for creating transgenic animals in many species and allows delivery of the desired material into the egg, whether it is RNA, DNA, or protein11. Initially, we tried transposon- and meganuclease-mediated approaches for integration of foreign DNA in the genome, but found in the course of the experiments that instead, random integration is a more efficient way for DNA incorporation in M. lignano. Random integration utilizes the molecular machinery of the host, integrating the provided DNA without the need for any additional components39. The method has its limitations, since the location and the number of integrated transgene copies cannot be controlled, and integration in a functional site can cause unpredictable disturbances and variation in transgene expression39. Indeed, we observed differences in the expression levels between independent transgenic lines for the EFA transgene reporter (Fig. 5).Transgene silencing might occur in a copy-dependent manner, as is the case in the germline of C. elegans40. However, the fact that we readily obtained transgenic lines with germline-specific expression (Fig. 3c–e) indicates that germline transgene silencing is not a major issue in M. lignano.
Variation of expression between different elongation factor 1 alpha transgenic lines. Fluorescence intensity is compared by taking images under the same exposure conditions at different exposure times (1.8 ms, 5 ms, 10 ms, and...
The efficiency of integration and germline transmission varied between 1 and 8% of injected eggs in our experiments (Table 1), which is reasonable, given that a skilled person can inject up to 50 eggs in 1 h. Although injection of a circular plasmid carrying a transgene can result in integration and germline transmission with acceptable efficiency (e.g., line NL23, Table 1), we found that injection of vector-free20 transgenes followed by ionizing irradiation of injected embryos with a dose of 2.5 Gy gave more consistent results (Table 1). Irradiation is routinely used in C. elegans for integration of extrachromosomal arrays, presumably by creating DNA breaks and inducing non-homologous recombination19. While irradiation can have deleterious consequences by inducing mutations, in our experiments we have not observed any obvious phenotypic deviations in the treated animals and their progeny. Nevertheless, for the downstream genetic analysis involving transgenic lines, several rounds of backcrossing to non-irradiated stock might be required to remove any introduced mutations, which is easily possible given that these worms are outcrossing and have a short generation time16,41. Despite the mentioned limitations, random integration of foreign DNA appears to be a straightforward and productive approach for generating transgenic lines in M. lignano and can be used as a basis for further development of more controlled transgenesis methods in this animal, including transposon-based42, integrase-based43, homology-based44, or CRISPR/Cas9-based45 approaches.
The draft genome assembly of the M. lignano DV1 line, which is also used in this study, was recently published9. The genome appeared to be difficult to assemble and even the 130× coverage of PacBio data resulted in the assembly with N50 of only 64 Kb9, while in other species N50 in the range of several megabases is usually achieved with such PacBio data coverages46. By adding Illumina and 454 data and using a different assembly algorithm, we have generated a substantially improved draft genome assembly, Mlig_3_7, with N50 scaffold size of 245.9 Kb (Table 2). The difficulties with the genome assembly stem from the unusually high fraction of simple repeats and transposable elements in the genome of M. lignano9. Furthermore, it was shown that M. lignano has a polymorphic karyotype and the DV1 line used for genome sequencing has additional large chromosomes (ref. 24 and Supplementary Fig. 3), which further complicates the assembly. The chromosome duplication also complicates genetic analysis and in particular gene knockout studies. To address these issues, we have established a different wild-type M. lignano line, NL10, from animals collected in the same geographical location as DV1 animals. The NL10 line appears to have no chromosomal duplications or they are present at a very low rate in the population, and its measured genome size is 500 Mb (Supplementary Fig. 3). While the majority of transgenic lines reported here are derived from the DV1 wild-type line, we observed similar transgenesis efficiency when using the NL10 line (Table 1, line NL24). Therefore, we suggest that NL10 line is a preferred line for future transgenesis applications in M. lignano.
To facilitate the selection of promoter regions for transgenic reporter constructs, we have generated Mlig_RNA_3_7 transcriptome assembly, which incorporates information from 5′- and 3′-specific RNA-seq libraries, as well as trans-splicing signals, to accurately define gene boundaries. We integrated genome assembly, annotation and expression data using the UCSC genome browser software (Supplementary Fig. 5, http://gb.macgenome.org). For genes tested in this study, the regions up to 2 kb upstream of the transcription start sites are sufficient to faithfully reflect tissue-specific expression patterns of these genes (Fig. 3), suggesting the preferential proximal location of gene regulatory elements, which will simplify analysis of gene regulation in M. lignano in the future.
In conclusion, we demonstrate that transgenic M. lignano animals can be generated with a reasonable success rate under a broad range of conditions, from circular and linear DNA fragments, with and without irradiation, as single and double reporters, and for multiple promoters, suggesting that the technique is robust. Similar to transgenesis in C. elegans, Drosophila and mouse, microinjection is the most critical part of the technique and requires skill that can be developed with practice. The generated genomic resources and the developed transgenesis approach provide a technological platform for harvesting the power of M. lignano as an experimental model organism for research on stem cells and regeneration.
M. lignano lines and cultures
The DV1 inbred M. lignano line used in this study was described previously9,24,47. The NL10 line was established from 5 animals collected near Lignano, Italy. Animals were cultured under laboratory conditions in plastic Petri dishes (Greiner), filled with nutrient enriched artificial sea water (Guillard’s f/2 medium). Worms were fed ad libitum on the unicellular diatom Nitzschia curvilineata (Heterokontophyta, Bacillariophyceae) (SAG, Göttingen, Germany). Climate chamber conditions were set on 20 °C with constant aeration, a 14/10 h day/night cycle.
Cloning of the elongation factor 1 alpha promoter
The M. lignano EFA promoter sequence was obtained by inverse PCR. Genomic DNA was isolated using a standard phenol-chloroform protocol; fully digested by XhoI and subsequently self-ligated overnight (1 ng/μl). Diluted self-ligated gDNA was used for inverse PCR using the EFA specific primers Efa_IvPCR_rv3 5′-TCTCGAACTTCCACAGAGCA-3′ and Efa_IvPCR_fw3 5′-CAAGAAGGAGGAGACCACCA-3′. Subsequently, nested PCR was performed using the second primer pair Efa_IvPCR_rv2 5′-AAGCTCCTGTGCCTCCTTCT-3′ and Efa_IvPCR_fw2 5′-AGGTCAAGTCCGTCGAAATG-3′. The obtained fragment was cloned into p-GEM-T and sequenced. Later on, the obtained sequence was confirmed with the available genome data. Finally, the obtained promoter sequence was cloned into two different plasmids: the MINOS plasmid (using EcoRI/NcoI) and the I-SceI plasmid (using PacI/AscI).
Highly expressed transcripts were identified from RNA-seq data8 and codon weight matrices were calculated using the 100 most abundantly expressed non-redundant genes. C. elegans Codon Adapter code48 was adapted for M. lignano (http://www.macgenome.org/codons) and used to design codon-optimized coding sequences (Supplementary Data 1). Gene fragments (IDT, USA) containing codon-optimized sequences, EFA 3′UTR and restriction cloning sites, were inserted into the pCS2+ vector to create optiMac plasmids used in the subsequent promoter cloning.
Cloning of tissue-specific promoters
Promoters were selected using Mlig_3_7, as well as several earlier M. lignano genome assemblies and MLRNA1509 transcriptome assembly8. RAMPAGE signal was used to identify the transcription start site and an upstream region of 1–2.5 kb was considered to contain the promoter sequence. An artificial ATG was introduced after the presumed transcription start site. This ATG was in-frame with the GFP of the target vector. The selected regions were cloned into optiMac vector using HindIII and BglII sites. Primers and cloned promoter sequences are provided in Supplementary Data 1.
Preparation and collection of eggs
Worms used for egg laying were kept in synchronized groups of roughly 500 per plate and transferred twice per week to prevent mixing with newly hatching offspring. The day before microinjections, around 1000 worms from 2 plates were combined (to increase the number of eggs laid per plate) and transferred to plates with fresh f/2 medium and no food (to remove the leftover food from the digestive tracks of the animals as food debris can attach to the eggs and impair the microinjections by clogging needles and sticking to holders). On the day of the injections, worms were once again transferred to fresh f/2 without food to remove any debris and eggs laid overnight. Worms were kept in the dark for 3 h and then transferred to light. After 30 min in the light, eggs were collected using plastic pickers made from microloader tips (Eppendorf, Germany), placed on a glass slide in a drop of f/2 and aligned in a line for easier handling.
Needles used in the microinjection procedure were freshly pulled using either borosilicate glass capillaries with filament (BF100-50-10, Sutter Instrument, USA) or aluminosilicate glass capillaries with filament (AF100-64-10, Sutter Instrument, USA) on a Sutter P-1000 micropipette puller (Sutter Instrument, USA) with the following settings: Heat = ramp-34, Pull = 50, Velocity = 70, Time = 200, Pressure = 460 for borosilicate glass and Heat = ramp, Pull = 60, Velocity = 60, Time = 250, Pressure = 500 for aluminosilicate glass. The tips of the needles were afterwards broken and sharpened using a MF-900 microforge (Narishige, Japan). Needles were loaded using either capillary motion or microloader tips (Eppendorf, Germany). Embryos were kept in position using glass holders pulled from borosilicate glass capillaries without a filament (B100-50-10, Sutter Instrument, USA) using P-1000 puller with the following settings: Heat = ramp + 18, Pull = 0, Velocity = 150, Time = 115, Pressure = 190. The holders were broken afterwards using a MF-900 microforge to create a tip of ~140 µm outer diameter and 50 µm inner diameter. Tips were heat-polished to create smooth edges and bent to a ~20° angle.
All microinjections were carried out on fresh one-cell stage M. lignano embryos. An AxioVert A1 inverted microscope (Carl Zeiss, Germany) equipped with a PatchMan NP2 for the holder and a TransferMan NK2 for the needle (Eppendorf, Germany) was used to perform all of the micromanipulations. A FemtoJet express (Eppendorf, Germany), with settings adjusted manually based on the amount of mucous and debris surrounding the embryos, was used as the pressure source for microinjections. A PiezoXpert (Eppendorf, Germany) was used to facilitate the penetration of the eggshell and the cell membrane of the embryo.
Irradiation was carried out using a IBL637 Caesium-137 source (CISbio International, France). Embryos were exposed to 2.5 Gy of γ-radiation within 1 h post injection.
Establishing transgenic lines
Positive hatchlings (P0) were selected based on the presence of fluorescence and transferred into single wells of a 24-well plate. They were then crossed with single-wild-type worms that were raised in the same conditions. The pairs were transferred to fresh food every 2 weeks. Positive F1 animals from the same P0 cross were put together on fresh food and allowed to generate F2 progeny. After the population of positive F2 progeny grew to over 200 hatchlings, transgenic worms were singled out and moved to a 24-well plate. The selected worms were then individually back-crossed with wild-type worms to distinguish F2 animals homozygous and heterozygous for the transgene. The transgenic F2 worms that gave only positive progeny in the back-cross (at least 10 progeny observed) were assumed to be homozygous, singled out, moved to fresh food and allowed to lay eggs for another month to purge whatever remaining wild-type sperm from the back-cross. After the homozygous F2 animals stopped producing new offspring, they were crossed to each other to establish a new transgenic line. The lines were named according to guidelines established at http://www.macgenome.org/nomenclature.html.
Images were taken using a Zeiss Axio Zoom V16 microscope with an HRm digital camera and Zeiss filter sets 38HE (FITC) and 43HE (DsRed), an Axio Scope A1 with a MRc5 digital camera or an Axio Imager M2 with an MRm digital camera.
Southern blot analysis
Southern blots were done using the DIG-System (Roche), according to the manufacturer’s manual with the following parameters: vacuum transfer at 5 Hg onto positively charged nylon membrane for 2 h, UV cross-linking 0.14 J/cm2, overnight hybridization at 68 °C.
Identification of transgene integration sites
The Universal GenomeWalker 2.0 Kit (Clontech Laboratories, USA) with restriction enzymes StuI and BamHI was used according to the manufacturer’s protocol. Sanger sequencing of PCR products was performed by GATC Biotech (Germany).
Whole mount in situ hybridization
cDNA synthesis was carried out using the SuperScript III First-Strand Synthesis System (Life Technologies, USA), following the protocol supplied by the manufacturer. Two micrograms of total RNA were used as a template for both reactions: one with oligo(dT) primers and one with hexamer random primers. Amplification of selected DNA templates for ISH probes was performed by standard PCR with GoTaq Flexi DNA Polymerase (Promega, USA). Amplified fragments were cloned into pGEM-T vector system (Promega, USA) and validated by Sanger sequencing. Primers used for amplification are listed in Supplementary Data 1. Templates for riboprobes were amplified from sequenced plasmids using High Fidelity Pfu polymerase (Thermo Scientific, USA). pGEM-T backbone binding primers: forward (5′-CGGCCGCCATGGCCGCGGGA-3′) and reversed (5′-TGCAGGCGGCCGCACTAGTG-3′) and versions of the same primers with an upstream T7 promoter sequence (5′-GGATCCTAATACGACTCACTATAGG-3′. Based on the orientation of the insert in the vector either forward primer with T7 promoter and reverse without or vice versa, were used to amplify ISH probe templates. Digoxigenin (DIG) labeled RNA probe synthesis was performed using the DIG RNA labeling Mix (Roche, Switzerland) and T7 RNA polymerase (Promega, USA) following the manufacturer protocol. The concentration of all probes was assessed with the Qubit RNA BR assay (Invitrogen). Probes were then diluted in Hybridization Mix49 (20 ng/µl), and stored at −80 °C. The final concentration of the probe and optimal hybridization temperature were optimized for every probe separately. Whole mount in situ hybridization was performed following a published protocol49. Pictures were taken using a standard light microscope with DIC optics and an AxioCam HRC (Zeiss, Germany) digital camera.
DV1 and NL10 worms were cut above the testes and left to regenerate for 48 h to increase the amount of dividing cells24. Head fragments were collected and treated with 0.2% colchicine in f/2 (Sigma, C9754-100 mg) for 4 h at 20 °C to arrest cells in mitotic phase. Head fragments were then collected and treated with 0.2% KCl as hypotonic treatment for 1 h at room temperature. Fragments were then put on SuperfrostPlus slides (Fisher, 10149870) and macerated using glass pipettes while being in Fix 1 solution (H2O: EtOH: glacial acetic acid 4:3:3). The cells were then fixed by treatment with Fix 2 solution (EtOH: glacial acetic acid 1:1) followed by Fix 3 solution (100% glacial acetic acid), before mounting by using Vectashield with Dapi (Vectorlabs, H-1200). At least three karyotypes were observed per worm and 20 worms were analyzed per line.
Genome size measurements
Genome size of the DV1 and NL10 lines was determined using flow cytometry approach50. In order eliminate the residual diatoms present in the gut, animals were starved for 24 h. For each sample 100 worms were collected in an Eppendorf tube. Excess f/2 was aspirated and worms were macerated in 200 µl 1× Accutase (Sigma, A6964-100ML) at room temperature for 30 min, followed by tissue homogenization through pipetting. 800 µl f/2 was added to the suspension and cells were pelleted by centrifugation at 4 °C, 1000 r.p.m., 5 min. The supernatant was aspirated and the cell pellet was resuspended in the nuclei isolation buffer (100 mM Tris-HCl pH 7.4, 154 mM NaCl, 1 mM CaCl2, 0.5 mM MgCl2, 0.2% BSA, 0.1% NP-40 in MilliQ water). The cell suspension was passed through a 35 µm pore size filter (Corning, 352235) and treated with RNase A and 10 mg/ml PI for 15 min prior to measurement. Drosophila S2 cells (gift from O. Sibon lab) and chicken erythrocyte nuclei (CEN, BioSure, 1006, genome size 2.5 pg) were included as references. The S2 cells were treated in the same way as Macrostomum cells. The CEN were resuspended in PI staining buffer (50 mg/ml PI, 0.6% NP-40 in calcium and magnesium free Dulbecco’s PBS Life Technologies, 14190136). Fluorescence was measured on a BD FacsCanto II Cell Analyzer first separately for all samples and then samples were combined based on the amount of cells to obtain an even distribution of different species. The combined samples were re-measured and genome sizes calculated using CEN as a reference and S2 as positive controls (Supplementary Fig. 3).
Preparation of genomic libraries
One week prior to DNA isolation animals were kept on antibiotic-containing medium. Medium was changed every day with 50 μg/ml streptomycin or ampicillin added in alternating fashion. Worms were starved 24 h prior to extraction, and then rinsed in fresh medium. Genomic DNA was extracted using the USB PrepEase Genomic DNA Isolation kit (USB-Affymetrix, Cat. No. 78855) according to manufacturer’s instructions. For the lysis step worms were kept in the supplied lysis buffer (with Proteinase K added) at 55 °C for 30–40 min and mixed by inverting the tube every 5 min. DNA was ethanol-precipitated once following the extraction and resuspended in TE buffer (for making 454 libraries Qiagen EB buffer was used instead). Concentration of DNA samples was measured with the Qubit dsDNA BR assay kit (Life Technologies, Cat. No. Q32850).
454 shotgun DNA libraries were made with the GS FLX Titanium General Library Preparation Kit (Roche, Cat. No. 05233747001), and for paired-end libraries the set of GS FLX Titanium Library Paired-End Adaptors (Roche, Cat. No. 05463343001) was used additionally. All the libraries were made following the manufacturer’s protocol and sequenced on 454 FLX and Titanium systems.
Illumina paired-end genomic libraries were made with the TruSeq DNA PCR-free Library Preparation Kit (Ilumina, Cat. No. FC-121-3001) following the manufacturer’s protocol. Long-range mate-pair libraries were prepared with the Nextera Mate Pair Sample Preparation Kit (Illumina, Cat. No. FC-132-1001) according to manufacturer’s protocol. Libraries were sequenced on the Illumina HiSeq 2500 system.
PacBio data (acc. SRX1063031) were assembled with Canu25 v. 1.4 with default parameters, except the errorRate was set to 0.04. The resulting assembly was polished with Pilon51 v. 1.20 using Illumina shotgun data mapped by Bowtie52 v. 2.2.9 and RNA-seq data mapped by STAR53 v. 2.5.2b. Next, scaffolding was performed by SSPACE26 v. 3.0 using paired-end and mate-pair Illumina and 454 data. Mitochondrial genome of M. lignano was assembled separately from raw Illumina reads using the MITObim software54 and the Dugesia japonica complete mitochondrial genome (acc. NC_016439.1) as a reference. The assembled mitochondrial genome differed from the recently published M. lignano mitochondrial genome55 (acc. no. MF078637) in just 1 nucleotide in an intergenic spacer region. The genome assembly scaffolds containing mitochondrial sequences were filtered out and replaced with the separately assembled mitochondrial genome sequence. The final assembly was named Mlig_3_7. Genome assembly evaluation was performed with REAPR27 and FRCbam28 software using HUB1_300 paired-end library and DV1-6kb-1, HUB1-3_6 kb, HUB1-3_7 kb, ML_8KB_1 and ML_8KB_2 mate-pair libraries (Supplementary Table 3).
Previously published M. lignano RNA-seq data8,31 (SRP082513, SRR2682326) and the de novo transcriptome assembly MLRNA150904 (ref. 8) were used to generate an improved genome-guided transcriptome assembly. First, trans-splicing and polyA-tail sequences were trimmed from MLRNA150904 and the trimmed transcriptome was mapped to the Mlig_3_7 genome assembly by BLAT56 v. 36 × 2 and hits were filtered using the pslCDnaFilter tool with the parameters “-ignoreNs -minId = 0.8 -globalNearBest = 0.01 -minCover = 0.95 –bestOverlap”. Next, RNA-seq data were mapped to genome by STAR53 v. 2.5.2b with parameters “--alignEndsType EndToEnd --twopassMode Basic --outFilterMultimapNmax 1000”. The resulting bam files were provided to StringTie29 v. 1.3.3 with the parameter “--rf”, and the output was filtered to exclude lowly expressed antisense transcripts by comparing transcripts originating from the opposite strands of the same genomic coordinates and discarding those from the lower-expressing strand (at least fivefold read count difference). The filtered StringTie transcripts were merged with the MLRNA150904 transcriptome mappings using meta-assembler TACO30 with parameters “--no-assemble-unstranded --gtf-expr-attr RPKM --filter-min-expr 0.01 --isoform-frac 0.75 --filter-min-length 100” and novel transcripts with RPKM <0.5 and not overlapping with MLRNA150904 mappings were discarded. The resulting assembled transcripts were termed ‘Transcriptional Units’ and the assembly named Mlig_RNA_3_7_DV1.v1.TU. To reflect closely related transcripts in their names, sequences were clustered using cd-hit-est from the CD-HIT v. 4.6.1 package57 with the parameters “-r 0 -c 0.95 -T 0 -M 0”, and clustered transcripts were given the same prefix name. Close examination of the transcriptional units revealed that they often represented precursor mRNA for trans-splicing and contained several genes. Therefore, further processing of the transcriptional units to identified boundaries of the encoded genes was required. For this, we developed computational pipeline TBONE (Transcript Boundaries based ON experimental Evidence), which utilizes exclusively experimental data to determine precise 5′ and 3′ ends of trans-spliced mRNAs. Raw RNA-seq data were parsed to identify reads containing trans-splicing sequences, which were trimmed, and the trimmed reads were mapped to the genome assembly using STAR53. The resulting wiggle files were used to identify signal peaks corresponding to sites of trans-splicing. Similarly, for the identification of polyadenylation sites we used data generated previously8 with CEL-seq library construction protocol and T-fill sequencing method. All reads originating from such an approach correspond to sequences immediately upstream of poly(A) tails and provide exact information on 3′UTR ends of mRNAs. The generated trans-splicing and poly(A) signals were overlapped with genomic coordinates of transcriptional units by TBONE, ‘cutting’ transcriptional units into processed mRNAs with exact gene boundaries, where such experimental evidence was available. Finally, coding potential of the resulting genes was estimated by TransDecoder58, and transcripts containing ORFs but missing a poly(A) signal and followed by transcripts without predicted ORF but with poly(A) signal were merged if the distance between the transcripts was not >10 kb and the spanning region was repetitive. The resulting assembly was named Mlig_RNA_3_7_DV1.v1.genes and includes alternatively spliced and non-coding transcripts. To comply with strict requirements for submission of genome annotations to DDBJ/ENA/GenBank, the transcriptome was further filtered to remove alternative transcripts with identical CDS, and to exclude non-coding transcripts and transcripts overlapping repeat annotations. This final transcriptome assembly was named Mlig_RNA_3_7_DV1.v1.coregenes and used in annotation of the Mlig_3_7 genome assembly for submission to DDBJ/ENA/GenBank.
Annotation of transposable elements and genomic duplications
Two methods were applied to identify repetitive elements de novo both from the raw sequencing data and from the assembled scaffolds. Tedna software59 v. 1.2.1 was used to assemble transposable element models directly from the repeated fraction of raw Illumina paired-end sequencing reads with the parameters “-k 31 -i 300 -m 200 -t 37 --big-graph = 1000”. To mine repeat models directly from the genome assembly, RepeatModeler package (http://www.repeatmasker.org) was used with the default settings. Identified repeats from both libraries were automatically annotated using RepeatClassifier perl script from the RepeatModeler package against annotated repeats represented in the Repbase Update – RepeatMasker edition database60 v. 20170127. Short (<200 bp) and unclassified elements were filtered out from both libraries. Additional specific de novo screening for full-length long terminal repeats (LTR) retrotransposons was performed using the LTRharvest tool61 with settings “-seed 100 -minlenltr 100 -maxlenltr 3000 -motif tgca -mindistltr 1000 -maxdistltr 20000 -similar 85.0 -mintsd 5 -maxtsd 20 -motifmis 0 -overlaps all”. Identified LTR retrotransposons were then classified using the RepeatClassifier perl script filtering unclassified elements. Generated repeat libraries were merged together with the RepeatMasker60 library v. 20170127. The resulted joint library was mapped on the genome assembly with RepeatMasker. Tandem repeats were annotated and masked with Tandem Repeat Finder
Adult stem cells are proposed to have acquired special features to prevent an accumulation of DNA-replication errors. Two such mechanisms, frequently suggested to serve this goal are cellular quiescence, and non-random segregation of DNA strands during stem cell division, a theory designated as the immortal strand hypothesis. To date, it has been difficult to test the in vivo relevance of both mechanisms in stem cell systems. It has been shown that in the flatworm Macrostomum lignano pluripotent stem cells (neoblasts) are present in adult animals. We sought to address by which means M. lignano neoblasts protect themselves against the accumulation of genomic errors, by studying the exact mode of DNA-segregation during their division.
In this study, we demonstrated four lines of in vivo evidence in favor of cellular quiescence. Firstly, performing BrdU pulse-chase experiments, we localized ‘Label-Retaining Cells’ (LRCs). Secondly, EDU pulse-chase combined with Vasa labeling demonstrated the presence of neoblasts among the LRCs, while the majority of LRCs were differentiated cells.We showed that stem cells lose their label at a slow rate, indicating cellular quiescence. Thirdly, CldU/IdU− double labeling studies confirmed that label-retaining stem cells showed low proliferative activity. Finally, the use of the actin inhibitor, cytochalasin D, unequivocally demonstrated random segregation of DNA-strands in LRCs.
Altogether, our data unambiguously demonstrated that the majority of neoblasts in M. lignano distribute their DNA randomly during cell division, and that label-retention is a direct result of cellular quiescence, rather than a sign of co-segregation of labeled strands.
Citation: Verdoodt F, Willems M, Mouton S, De Mulder K, Bert W, Houthoofd W, et al. (2012) Stem Cells Propagate Their DNA by Random Segregation in the Flatworm Macrostomum lignano. PLoS ONE 7(1): e30227. https://doi.org/10.1371/journal.pone.0030227
Editor: Sebastian D. Fugmann, National Institute on Aging, United States of America
Received: December 23, 2010; Accepted: December 14, 2011; Published: January 19, 2012
Copyright: © 2012 Verdoodt et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported by an IWT doctoral grant (Institute for the Promotion of Innovation through Science and Technology in Flanders, IWT-Vlaanderen) to FV, MW and SM, and by FWF grant 18099 to PL (Austria). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Adult stem cells (ASCs) have a long-term and dual responsibility to both self-renew and produce differentiated progeny, thereby playing a crucial role during the entire lifetime of an organism , . Given the constant demand for proliferation and the error-prone nature of DNA replication, these cells possess a high risk for malignant transformation . As a consequence, it has long been postulated that ASCs might have acquired specialized features to protect their genome , . A highly efficient DNA-repair system is commonly described as a stem cell trait, which would serve this purpose .
Additionally, a putative mechanism by which ASCs might limit accumulating erroneous genetic information, was originally proposed by Cairns  as the immortal strand hypothesis. According to this hypothesis, stem cells segregate their DNA strands non-randomly upon asymmetric self-renewing cell divisions. Those sister chromatids containing the original template DNA strands are selectively retained in one daughter cell, destined to be the renewed stem cell. The newly synthesized strands, which might have acquired mutations during replication, are passed on to the tissue committed cell. A common strategy to verify this hypothesis, relies on pulse-chase studies with nucleotide tracers, such as tritiated thymidine, bromodeoxyuridine (BrdU), or chlorodeoxyuridine (CldU). Labeling the original ‘immortal’ DNA strands when they are synthesized during development or regeneration, should result in ‘Label-Retaining Cells’ (LRCs), considering that these labeled strands are co-segregated during cell divisions (Figure 1A, top panel).
Figure 1. Possible interpretation of label-retention studies, using a double labeling approach.
(A): Cairns' theory, known as the immortal strand hypothesis, postulates that adult stem cells (ASCs) segregate their DNA-strands non-randomly and permanently retain original template DNA. Using a thymidine derivate such as CldU, these template DNA strands can be labeled at the moment they are synthesized during development, which results in two daughter cells in which complementary DNA-strands are composed of 1 labeled template strand next to an unlabeled normal DNA strand. Because of co-segregation of the labeled template DNA-strands, from the second division after establishment of labeled immortal strands on, the label is passed on to only one daughter cell. Therefore, cells are able to retain label indefinitely during adulthood and are referred to as label-retaining cells (LRCs). By performing a second pulse with another thymidine analog such as IdU, LRCs can become double labeled. (B): If DNA is segregated randomly, labeled DNA-strands which are created during a first pulse period with CldU, are distributed over both daughter cells, instead of only one. By consequence, in a regularly cycling cell, the label is diluted under the detection threshold after a certain number of cell cycles in CldU-free medium (left panel). Thus, initially-labeled cells do not retain the label and a second pulse-period with IdU will not result in double labeled cells. However, if cells remain quiescent after they incorporated CldU, the chance of dilution of the label is reduced due to low or even absent cell proliferation (right panel). This results in LRCs. Creating double labeled LRCs after a second pulse with a thymidine derivate (IdU), is therefore possible, yet unlikely because of low cell cycle activity. Abbreviations: LRC, label-retaining cell; CldU, 5-chloro-2′-deoxyuridine; IdU, 5-iodo-2′-deoxyuridine.
Alternatively, retention of label in stem cells can likewise be explained as a result of cellular quiescence. Restricting the number of stem cell divisions seems an equally valuable mechanism for preservation of genome integrity and furthermore prevents stem cell exhaustion –. Low or absent proliferative activity, after cells were labeled with nucleotide tracers, reduces the chance of label-dilution and allows quiescent ASCs to be identified as ‘Label-Retaining Cells’ (LRCs) (Figure 1B, top right panel). Conversely, in more rapidly cycling progeny cells the label is gradually diluted (Figure 1B, top left panel). Performing a double labeling protocol using a second nucleotide tracer serves as a promising tool to assess information on the proliferative activity of LRCs (Figure 1A, B, bottom panel).
Elucidating the label-retention theory remains a matter of intense debate, fueled by publications confirming the theory of cellular quiescence on one hand , –, versus those supporting non-random segregation of DNA strands on the other hand –. It has been shown that culture environments can alter the patterning of cells in ways that modify their fates and proliferative potential , . Therefore, the use of model organisms in which stem cells can be studied in vivo has attracted substantial attention –. However, the in vivo data on this topic is mainly gathered in systems in which the analysis of stem cell behavior is hindered by the rare incidence of stem cells, relative inaccessibility of these cells for experimental manipulation in vivo, and lack of specific stem cell markers .
Over the last decennia, flatworms have been put forward as valuable model organisms to unravel the complex biology of stem cells –. These simple, triploblastic metazoans exhibit a powerful stem cell system that is maintained through adult life and which lies at the root of their exceptional developmental plasticity and regeneration capacity . The flatworm stem cell population is comprised of pluripotent stem cells, referred to as neoblasts, which remain mitotically active during adulthood, unlike all differentiated cells in the organism –. Among flatworms, Macrostomum lignano (Figure 2A) has been recently described as a highly advantageous model for in vivo stem cell research , –. Advantages are the ease of culturing , , the short embryonic and post-embryonic development (5 and 14 days, respectively), and the limited number of cells (25 000 in total) which facilitates cell quantification . Furthermore, neoblasts are well characterized and present in large numbers (6.5% of the total cell number) . They can easily be distinguished from non-stem cells, based on morphological traits, and by using using an antibody against neoblast-specific Macvasa proteins , . Immunohistochemical staining of S-phase neoblasts with the thymidine analog bromodeoxyuridine (BrdU), and mitotic neoblasts with an anti-phospho histone H3 mitosis marker (anti-phos-H3), have revealed a bilateral distribution of these cells ,  (Figure 2B). Pulse and pulse-chase studies with thymidine analogs such as BrdU can easily be performed by soaking the animals in the analog-containing medium during the pulse period. Moreover, an in vivo double labeling technique using two different thymidine derivates, iododeoxyuridine (IdU) and chlorodeoxyuridine (CldU), can be applied . To our knowledge, this technique has been performed only once before to test the segregation mode of DNA-strands in vivo. Altogether, these advantages enable in vivo analysis of the exact mode of DNA segregation in ASCs in the flatworm M. lignano.
Figure 2. Macrostomum lignano (Platyhelminthes).
(A): Light microscopic picture of an adult specimen, dorsal view (left panel). Schematic drawing (right panel). Abbreviations: R, rostrum; B, brain; E, eye; P, pharynx; G, gut; T, testis; O, ovary; D, developing egg. (B): Confocal projection of a double BrdU/phospho histone H3 immunostaining (green S-phase cells, red mitoses converted to magenta) after a 30-min BrdU pulse (no chase) in an adult animal. During homeostasis, proliferating neoblasts are distributed in a bilateral pattern. S-phase, nor mitotic cells are visible anterior to the eyes. Arrow indicates the level of the eyes. Anterior is on top. Scale bars: 50 µm.
We aimed to elucidate if label-retaining stem cells exist in M. lignano. Performing long-term pulse-chase studies, four different in vivo approaches were used. First, a single BrdU-pulse-chase experiment was performed to demonstrate the existence of LRCs. Second, among this population of LRCs, Macvasa-positive (Macvasa+) neoblasts were identified. Third, double labeling of the LRCs with chlorodeoxyuridine (CldU) and iododeoxyuridine (IdU) gave information on the proliferative activity. Finally, the actin inhibitor cytochalasin D was used to directly analyze the segregation of labeled DNA strands at the single-cell level. Altogether, our results demonstrate that in M. lignano random segregation of DNA strands is predominant, and that label-retention is a direct result of cellular quiescence.
Cultures of M. lignano were reared in standard culture medium (f/2)  as described previously , . To obtain animals of a standardized age, adult worms were put together for 24 hours, after which the eggs were collected. Animals that were pulsed with a thymidine analog, were protected from light.
BrdU-pulse labeling and immunocytochemistry in whole-mount organisms and macerated cell suspensions
A 24-hour incubation period in the thymidine analog 5-bromo-2′-deoxyuridine (BrdU - Sigma) was given to 11 standardized age groups of embryos (at day 1, 2, 3, 4, or 5 of development) and hatchlings (at day 6, 7, 8, 9, 10, or 11 of development). Together, the initial five age groups cover the embryonic development of M. lignano, while the following groups cover the first six days of post-embryonic development. Both embryos and hatchlings were pulsed, simply by soaking them in f/2 containing BrdU (500 µM). Animals were then kept in in standard culture medium, in the presence of food (ad libitum), for two or six months in the absence of BrdU. Subsequently, BrdU positive cells were localized using the protocol described below.
The procedure for visualization of the incorporated BrdU in whole mount animals was modified from a previous publication . Specimens were relaxed in MgCl2 (1∶1 MgCl2.6H2O (7.14%):f/2, 5 min – MgCl2.6H2O (7.14%), 5 min), fixed in 4% paraformaldehyde (PFA, 30 min), and rinsed in PBS-T (phosphate-buffered saline+0.1% Triton X-100, 3×10 min). Animals were then treated with Protease XIV (0.2 mg/ml in PBS-T, 37°C, under visual control) and DNA was denatured with 2 N HCl (1 h, 37°C). Rinsing with PBS-T (6×10 min) and blocking in BSA-T (PBS-T+1% bovine serum albumin, 30 min) were followed by overnight incubation in the primary antibody, rat-anti-BrdU (1∶800 in BSA-T, 4°C - AbD Serotec). Subsequently, animals were washed in PBS-T (3×10 min) and incubated in the secondary antibody, FITC-conjugated donkey-anti-rat (1∶600 in BSA-T, 1 h – Rockland). Finally animals were rinsed in PBS (3×10 min) and mounted in Vectashield (Vector Laboratories).
Simultaneous visualization of S- phase and mitotic neoblasts, using anti-BrdU and anti-phos-H3, was performed as described by Nimeth et al., with the exceptions that both the Protease XIV and HCl treatment was performed as described above. Rhodamine-conjugated goat-anti-rabbit (1∶150 in BSA-T, 1 h – Millipore) was used as a secondary antibody for the mitosis marker.
In macerated cell suspensions, the incorporated BrdU was visualized as described before , though some modifications were made. Twenty animals were incubated in 100 µl of maceration solution (glacial acetic acid∶glycerol∶distilled water 1∶1∶13 - 9% sucrose, 10 min), after which calcium/magnesium-free medium (CMF, 100 µl) was added. Thirty minutes after addition, animals were gently pipetted until they fell apart into single cells. Cells were then pelleted (130× g, 20 min), supernatant was removed, and the pellet was resuspended in PBS (200 µl). The cell suspension was spread onto poly-L-lysine coated slides. The staining of BrdU, was performed directly on these slides in a humid chamber, and was identical to the protocol for whole-mount preparations, with the exclusion of the Protease XIV step. Prior to the mounting of slides with Vectashield, DNA was stained using DAPI (1 µg/ml in PBS, 1 h). The morphology of single cells was studied, following the methods described earlier , . Neoblasts were identified as small, rounded cells (5–10 µm) with a large nucleus and scanty cytoplasm.
Double labeling with CldU and IdU and immunocytochemistry of whole mounts
Standardized age groups of embryos and hatchlings were pulsed with the thymidine analog 5-chloro-2′-deoxyuridine (CldU; 500 µM in f/2, 24 h - Sigma) following the same protocol as described for BrdU-labeling. Animals were then chased for six months in the presence of food (ad libitum) in CldU-free standard culture medium, after which they were pulsed with 5-iodo-2′-deoxyuridine (IdU; 50 µM in f/2 – Sigma) continuously for 7 days. The following steps were identical to the single BrdU-labeling in whole-mount animals (starting from MgCl2-relaxation until incubation in BSA-T). Animals were then incubated in the first primary antibody, mouse-anti-IdU (1∶800 in BSA-T, overnight, 4°C - Becton-Dickinson); washed in PBS-T (3×10 min); incubated in stringency buffer (0.5 M NaCl+36 mM Tris HCl+0.5% Tween 20, 15 min) for removal of nonspecifically-bound primary antibody; and washed again in PBS-T (3×10 min). Subsequently, specimens were incubated in the first secondary antibody, Alexa Fluor 568-conjugated goat-anti-mouse (1∶900 in BSA-T, 1 h - Invitrogen); the second primary antibody, rat-anti-CldU (1∶800 in BSA-T, overnight, 4°C - AbD Serotec); and the second secondary antibody, FITC-conjugated donkey-anti-rat (1∶600 in BSA-T, 1 h - Rockland). Incubation-periods in antibodies were separated by washing steps in PBS-T (3×10 min). Finally, animals were rinsed in PBS (3×10 min) and mounted in Vectashield.
EdU-pulse labeling, immunocytochemistry in macerated cell suspensions and the use of cytochalasin D
Embryos and hatchlings, standardized by age, were soaked in the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU; 20 µM in f/2 - Invitrogen) during development, respectively from day 1 until day 5 continuously, and from day 6 until day 11 continuously. Animals were then chased in the presence of food (ad libitum) for two months in EdU-free medium, followed by a seven-day incubation period in the actin-binding protein cytochalasin D (5 µM in f/2 - Sigma). Subsequently, they were macerated, following the protocol described earlier and cells were spread onto poly-L-lysine coated slides, washed in PBS (3×10 min), blocked with BSA-T-1% (PBS+1% Triton X-100+1% BSA, overnight, 4°C) and incubated in Click-iT® EdU reaction cocktail (concentrations according to manufacturer's instructions - Invitrogen). Afterwards, slides were washed thoroughly in BSA-T-1% (1 h), DNA was stained with DAPI (1 µg/ml in PBS, 1 h) and cells were mounted using Vectashield (Vector Laboratories).
EdU-pulse labeling, immunocytochemistry in whole mounts and the use of Macvasa antibody
An EdU-pulse was performed as described above, in embryos (day 1–5, continuously) and hatchlings (day 6–11, continuously), after which a three-month chase was performed in the presence of food in EdU-free medium. Subsequently, animals were relaxed, fixed, and rinsed with PBS-T, as described above. Blocking was performed with BSA-T (2 h), followed by incubation in Click-iT® EdU reaction cocktail (concentrations according to manufacturer's instructions – Invitrogen). Next, Macvasa+ cells were visualized as described by Pfister et al., using primary rabbit-anti-Macvasa and secondary TRITC-conjugated goat-anti-rabbit. Finally, animals were rinsed in PBS (3×10 min) and mounted in Vectashield.
Imaging and quantification of LRCs in whole-mounts
Epifluorescence and phase-contrast microscopy was performed on a Zeiss Axiovert 200 M inverted microscope, followed by image processing using AxioVision 4.7.2. software (Zeiss) and Photoshop CS2. A Nikon Eclipse C1si confocal microscope was used for generating confocal images of whole mount animals. An argon laser (488 nm) in combination with a narrow band-pass filter (BP 515/30), and a helium-neon laser (543 nm) in combination with a narrow band-pass filter (BP 593/40) were used for visualization of the FITC-fluorochromes (CldU) and the Alexa Fluor 568-fluorochromes (IdU), respectively. Images where processed, using Nikon EZ-C1 3.40 software and Adobe Photoshop CS2.
Quantification of BrdU+ LRCs was performed on confocal images, using the free software program Image J . Images were prepared by performing automatic thresholding, after which cells were quantified automatically, using the ‘Analyze Particles’ plug-in in Image J. In order to exclude labeled differentiated cells from the cell counts, exclusion parameters were activated based on size and shape of the labeled particles (Size pixel ∧2: 10–100; Circularity: 0.80–1.00). Based on their location in regions which are known to lack neoblasts in M. lignano, labeled cells in the rostrum (anterior to the eyes) and at the median axis were not considered to be stem cells and were therefore excluded from counts.
Statistical analysis was performed using Mann-Whitney U (BrdU+ LRC's) and Kruskal-Wallis (CldU+/IdU+ LRC's) tests.
Establishment of LRCs in M. lignano
To evaluate whether LRCs were present in M. lignano, animals were pulsed with BrdU during development, allowing nascent neoblasts to incorporate the thymidine analog into their DNA. For BrdU-incorporation, five groups of embryos and six groups of hatchlings, standardized by age, were pulsed at successive 24-hour time intervals during both embryonic (days 1–5) and postembryonic (days 6–11) development (Figure 3A). This wide developmental window was chosen to ascertain that the potential founder label retaining neoblasts were covered by the pulse period. In order to pinpoint a specific time interval during which these neoblasts originate, BrdU-incubation was limited to intervals of 24 hours. Following the BrdU-pulse, specimens were chased in the presence of food, for two and six months in BrdU-free medium. Subsequently, after both chase periods, 10 randomly chosen animals of every pulse-group (days 1–11) were sacrificed, BrdU was visualized, and animals were examined for the presence of LRC's.
Figure 3. LRCs can be established during both embryonic and post-embryonic development, and lose their label at a slow rate.
(A): Scheme of the experimental set-up. Animals were pulsed with BrdU (24 h) at successive 24-hour time frames of embryonic and post-embryonic development, followed by chase times of 2 and 6 months in BrdU-free medium. Subsequently BrdU was visualized and the presence of LRCs was analyzed. (B): Visualization of LRCs (green) in whole mount animals (confocal projections of BrdU immunostaining). Left panel, from left to right: animal pulsed at day 3 (ED), animal pulsed at day 6 (PED); both animals were chased for 2 months. Right panel, from left to right: animal pulsed at day 4 (ED), animal pulsed at day 6 (PED); both animals were chased for 6 months. Abbreviations: b., cluster of BrdU+ cells at the level of the brain; ph., cluster of BrdU+ cells at the mouth-pharynx region. Anterior is on top. (C): Graph representing the quantification of LRCs (mean number+Standard deviation, n = 22×5), chased for 2 and 6 months. LRCs were present in all animals of all pulse groups, both after two and six months. In animals, chased for 6 months, a significant lower number of LRCs was present, compared to animals that were pulsed for 2 months. Abbreviations: BrdU, 5-bromo-2′-deoxyuridine; ED, embryonic development; PED, post-embryonic development. (D): Visualization of LRCs in macerated cell suspensions. Superimposition of interference contrast and fluorescence images of BrdU immunostaining (left), interference contrast images (middle) and fluorescent images (right). Animals were pulsed during embryonic and post-embryonic development, and chased for 2 months. Pictures show labeled neoblasts with a large nucleus surrounded by a small rim of cytoplasm (C1, C2), and a labeled nerve cell (C3). Scale bars: B, 50 µm; D, 5 µm.
After two months of chase, the presence of cells that had retained the BrdU label was confirmed in all studied animals that were pulsed during embryonic and post-empbryonic development (total n = 110). LRCs were distributed all over the body (Figure 3B, left panel). A high density of BrdU+ cells was observed in a bilateral pattern, which is in accordance with the distribution of neoblasts in M. lignano (Figure 2B) . Outside this bilateral pattern, two separate clusters of labeled cells were found; one at the level of the brain, and another at the level of the mouth and pharynx. To test whether LRCs could be established during homeostasis as well, adult individuals were pulsed with BrdU and then chased for two months. Similarly, this resulted in the presence of LRCs in all studied individuals (data not shown).
After a six-month chase-period, 95% of a total of 110 randomly chosen animals could be labeled (n = 104), and LRCs were present in all of them. These labeled cells were scattered throughout the body (Figure 3B, right panel). An accumulation of labeled cells, similar to those after two months chase, at the brain region, the mouth-pharynx region, or both was visible in 75% of the animals.
For every pulse-group, both at two and six months chase, the number of LRCs was quantified in five animals (Figure 3C), as described in the Methods section. The number of LRCs was significantly lower in all pulse-groups at six months chase when compared to the same groups after two months chase (for all pulse-groups, p<0.05). Thus, a considerable amount of LRCs have lost their label over time, meaning that these cells are not able to retain label indefinitely, or labeled cells were replaced by the progeny of unlabeled neoblasts during tissue homeostasis. The number of LRCs after two months chase seemed to vary, with a mean value of 31 LRCs per animal (±9, n = 55). After 6 months chase, the mean number of LRCs for all pulse groups combined was 13 per animal (±6, n = 55).
These LRCs might represent differentiated progeny of labeled stem cells, in which case the label is retained due to the post-mitotic state of differentiated cells in flatworms. In order to verify whether LRCs, or a fraction thereof, could be identified as neoblasts, two month chased animals were macerated into single cells and BrdU was visualized. By analyzing the morphology of BrdU-positive cells, the existence of label-retaining neoblasts was confirmed (Figure 3D1,2). In addition, several differentiated labeled cells were found, including cells displaying the morphology of nerve cells (Figure 3D3) and epidermis cells (not shown).
Label-retaining stem cells are positive for the neoblast marker Macvasa
An additional experiment was performed to test whether neoblasts could be identified within the population of LRCs. For this purpose, a polyclonal antibody against a homolog of the highly conserved Vasa protein of M. lignano (Macvasa) was used. Unlike in other metazoans, where Vasa is almost exclusively detected in germ line cells, Macvasa in M. lignano is also present in a subset of somatic stem cells in a characteristic pattern - a ring of Macvasa-labeled spots of nuage surrounding the nucleus . Consequently, Macvasa can be used as a neoblast marker in this flatworm. In this experiment, LRCs were established using EdU, since HCl-denaturation is unnecessary for the visualization of this thymidine analog, which enabled simultaneous labeling of Macvasa proteins. The specificity of EdU-labeling was observed comparable to BrdU (see Figure S1, Text S1).
In the first pulse group, individuals were pulsed continuously with the thymidine analog EdU for five days during embryonic development (day 1 until day 5). A second group of individuals was pulsed continuously with EdU during the first six days of post-embryonic development (day 6 until day 11). Individuals were then chased for three months, in the presence of food (Figure 4A), after which EdU-positive (EdU+) and Macvasa+ cells were visualized.
Figure 4. Identification of label-retaining neoblasts (Macvasa).
Identification of neoblasts among the population of LRCs, using an antibody against neoblast-specific Macvasa proteins. (A): Scheme of the experimental set-up. Animals were pulsed continuously with EdU during embryonic development (day1–day5) and during post-embryonic development (day6–day11), followed by a chase time of 3 months in EdU-free medium. Subsequently, EdU was visualized in combination with Macvasa. (B): Visualization of label-retaining neoblasts in whole-mount animals (confocal image). LRCs (EdU) are green. The TRITC-signal of the Macvasa proteins, was converted to magenta. EdU+/Macvasa+ cell in a control animal (no chase) displays Macvasa proteins in a ring of nuage around the nucleus (upper left panel). EdU+/Macvasa+ cell in an animal pulsed during ED (upper right panel). EdU+/Macvasa+ cells in an animal pulsed during PED (lower left and right panel). LRCs that are Macvasa-negative (asterisks) are visible in individuals pulsed during ED and PED (right panels). Abbreviations: EdU, 5-ethynyl-2′-deoxyuridine; Vasa, Macvasa; ED, embryonic development; PED, post-embryonic development. Scale bars: 10 µm.
Macvasa+ LRCs were identified in all individuals pulsed during embryonic (day 1–day 5) as well as in individuals pulsed during post-embryonic development (day 6–day 11) (Figure 4B). Macvasa protein in double labeled cells, was visible in spots of nuage around the EdU-labeled nucleus, as observed previously .
Macvasa+ LRCs were located at the lateral sides of the animal, the area described to contain somatic neoblasts . Double positive cells were never observed in the testes, nor ovaries.
In conclusion, these results directly confirm the existence of neoblasts among the population of LRCs, which are distributed among other somatic neoblasts.
Label-retaining stem cells manifest low proliferative activity
To further analyze the proliferative activity of label-retaining stem cells, a CldU/IdU double labeling method was applied to label the S-phase of stem cells after a six months chase time. In these experiments, CldU was administered continuously for 24 hours to different groups of embryos and hatchlings at successive time frames of embryonic (days 1–5) and post-embryonic (days 6–11) development, followed by a chase time of six months in CldU-free culture medium. Following the chase period, a second pulse with IdU was performed for seven days continuously to embrace all LRCs that proliferated during this week (Figure 5A). Immediately afterwards, animals were immunostained for CldU and IdU. Consequently, every CldU+ LRC going through S-phase during the second pulse period with IdU incorporates this thymidine analog as well and therefore becomes double labeled.
Figure 5. Low proliferative activity of label-retaining cells (LRCs).
Analysis of the proliferative activity of LRCs in M. lignano, performing a double labeling technique with the proliferation markers CldU and IdU. (A): Scheme of the experimental set-up. Animals were pulsed with CldU (24 h) at successive 24-hour time frames during embryonic and post-embryonic development, followed by a chase time of six months in CldU-free medium. Subsequently animals were pulsed for 7 days continuously with IdU, after which both markers were visualized and the presence of double labeled cells was analyzed. (B): Double labeled LRC located in the tail region (left inset), in a whole mount animal (confocal plane) that was pulsed with CldU (green) on day 1 of embryonic development, chased for six months, and pulsed again with IdU (red, converted to magenta). Every LRC (labeled during the first pulse) that proliferates during the second pulse will become overlabeled with IdU, and is CldU+/IdU+ (white, indicated with arrow). LRCs that do not proliferate during the second pulse are CldU+ (green) and cells which are proliferating during the second, but not the first pulse, are IdU+ (magenta). Abbreviations: CldU, 5-chloro-2′-deoxyuridine; IdU, 5-Iodo-2′-deoxyuridine. Scale bars: C, 50 µm; C inset, 20 µm.
The presence of proliferating LRCs (CldU+/IdU+ cells) was confirmed in representatives of every pulse group (day 1–11) (Figure 5B). Since neoblasts are the only somatic cells that are actively dividing in M. lignano, this directly proves that each 24-hour pulse period resulted in neoblasts that were able to retain their label for six months. Hence, no specific time-frame could be pinpointed for the establishment of proliferating label-retaining stem cells. The distribution of all CldU+/IdU+ cells was in accordance with the normal distribution of neoblasts , except for two cells that were located in the rostrum. These two cells were probably differentiated and migrated during the seven day administration of the second pulse.
A quantitative study of double labeled cells was performed in animals that were pulsed during embryonic development (days 1–5) and chased for six months. Overall, the number of CldU+/IdU+ cells was very low, with an observed maximum of three double labeled cells per worm (8% of all animals observed, n = 38). In most animals (58%) zero double labeled cells were quantified, and 24% and 11% of all observed animals had one and two CldU+/IdU+ cells, respectively. When analyzed for each of the five pulse groups, the mean numbers of double positive cells per worm did not significantly differ between the groups (p>0.8). This low number of double labeled cells indicated little proliferative activity among the label-retaining stem cells.
The use of cytochalasin D indicates random segregation of DNA strands
In order to directly test the segregation pattern of DNA strands in vivo, cytochalasin D was used. This actin binding protein blocks cytokinesis, while karyokinesis is unaffected, thereby maintaining one cell with two daughter nuclei.
In order to incorporate EdU in all cells, embryos were pulsed continuously during the whole embryogenesis (day 1 until day 5). In a second pulse-group, hatchlings were continuously treated with EdU from day 6 until day 11. Animals were then chased in the presence of food for two months and subsequently incubated in cytochalasin D for one week (Figure 6A). Immediately afterwards, animals were macerated into single cells and stained for EdU. As a consequence, each label-retaining stem cell that proliferated during this one week incubation-period was blocked, resulting in a binucleate cell. This made it possible to analyze the distribution of labeled strands in the daughter nuclei.
Figure 6. Random distribution of labeled DNA-strands among daughter nuclei of LRCs.
Analysis of the distribution of labeled DNA-strands among daughter nuclei of label-retaining cells (LRCs) on single cell level, using the actin-binding protein cytochalasin D. Cytochalasin D is an actin-binding protein that inhibits cytokinesis, while karyokinesis remains unaffected. Thereby, binucleate cells are created, which enables analysis of the distribution of DNA-strands among daughter nuclei on single cell level. (A): Scheme of the experimental set-up. Animals were pulsed continuously with EdU during embryonic development (day1–day5) and during post-embryonic development (day6–day11), followed by a chase time of 2 months in EdU-free medium. Subsequently animals were soaked in cytochalasin D for 7 days, EdU was visualized and DNA was stained with DAPI. (B): Visualization of binucleate LRCs in macerated cell suspensions. Fluorescence images of EdU (left), DAPI (middle), and interference contrast images (right) of binucleate LRCs, pulsed during embryonic (B1) and post-embryonic (B2, B3) development. Binucleate EdU+ cells display equivalent EdU-signal in both daughter nuclei. Abbreviations: EdU, 5-ethynyl-2′-deoxyuridine. Scale bars: 5 µm.
All EdU+ binucleate cells that were observed displayed an equivalent EdU-signal in both daughter nuclei (Figure 6B). No cells were found that contained a labeled nucleus next to an unlabeled one, or otherwise displayed evidence of unequal fluorescence distribution.
In M. lignano four different in vivo approaches were used to analyze the exact segregation mode of DNA-strands during stem cell division. None of these approaches produced evidence for non-random segregation of DNA-strands, and were therefore inconsistent with the immortal strand hypothesis. In contrast, our long-term label-retention analyses are rather a confirmation of the existence of a population of relatively quiescent stem cells.
BrdU pulse-chase experiments were performed to test whether LRCs can be established in M. lignano. In order to enable pinpointing the origin of LRCs to a specific developmental window, an elaborate pulse scheme was designed. The data demonstrated that LRCs could be established in all specimen pulsed during 11 different time periods of development. A similar pulse-chase experiment in adult worms also resulted in LRCs. Thus, our study demonstrates that LRCs can be established in M. lignano, not only during the complete duration of embryonic development (day1–day 5), but also during post-embryonic development (day 6–day 11) and even during adulthood. Of possible concern was that label retention is caused by an artifactual withdrawal from the cell cycle, caused by a possible deleterious effect of the incorporated thymidine analog. However, our double labeling experiment contradicts this hypothesis, since label-retaining neoblasts are observed to proliferate. Previously, a continuous BrdU pulse (50 µM) from hatching to maturity in M. lignano, has been observed to result in viable labeled sperm . Furthermore, other reports on the use of BrdU in M. lignano (with continuous pulse durations up to 14 d) have demonstrated no major effect on the dynamics of proliferating cells, since pulsing was not observed to affect morphology, animal behavior, cell cycle dynamics of fast cycling cells, differentiation of BrdU+ cells, and sperm production and differentiation , , , , . Based on the results presented in this study and in previous studies, we can conclude that the effect of analog incorporation is minimal and that label-retention is not caused by a cell cycle arrest. Another possible caveat of label-retention studies is that the label is retained due to the post-mitotic state of differentiated cells. Working with M. lignano, however, enables identification of stem cells, based on their morphology , expression of Macvasa, and their ability to incorporate a thymidine analog, as the only proliferating somatic cells . In this study, the presence of label-retaining neoblasts among the LRCs was proven in three ways: (1) labeled neoblasts were identified morphologically, (2) Macvasa proteins were demonstrated in a subset of LRCs, and (3) a small number of LRCs were observed to incorporate IdU in our double labeling experiment. These double labeled cells were found in all pulse groups, meaning that every pulse that was performed resulted in stem cells which retained label for extended periods of time. Thus, if label-retention would be a result of non-random segregation of labeled DNA-strands and ‘immortal’ strands do exist in M. lignano, our observations indicate that they would be synthesized continuously during embryonic and post-embryonic development, as well as during homeostasis. Serially creating new ‘immortal’ strands is totally incompatible with the purpose of the immortal strand hypothesis. On the other hand, the results of our analysis for the establishment of LRCs are compatible with the existence of a population of relatively quiescent stem cells. The inability to pinpoint the origin of LRCs to a specific developmental window, has previously been observed to be compatible with the existence of quiescent stem cell in mice .
Quantitative analysis of LRCs after two and six months chase demonstrates a significant decline in the number of LRCs. Still, neoblasts are observed to be able to retain label for as long as six months, a period equivalent to the median life span of M. lignano. Thus, the label of LRCs is lost at an extremely slow rate, indicating little cell proliferation, a sign for cellular quiescence. To directly test the proliferative activity of LRCs during homeostasis, in vivo CldU/IdU double labeling experiments were performed. The extremely low numbers of double labeled LRCs demonstrate little proliferation. Both our single and double labeling experiment, therefore, deliver strong arguments for the existence of a population of quiescent stem cells in M. lignano. The combined outcome of our single and double labeling experiment after prolonged chase times, and the implications thereof, are explained in Fig. 7A. These experiments have demonstrated the establishment of a population of LRCs, consisting of labeled neoblasts on one hand, and differentiated progeny of labeled neoblasts on the other hand. By pulsing with CldU during embryonic or post-embryonic development, cells become labeled in S-phase (Fig. 7A, left panel). During successive development and growth, these CldU-labeled cells proliferate and create labeled progeny. The labeled progeny then migrates and differentiates to participate in homeostasis. As a result of proliferation, migration and differentiation CldU-labeled cells are distributed throughout the whole animal with some clustering at the brain and pharynx (Fig. 7A, middle panel). CldU labeled stem cells that go through S-phase during a second, 7 d-pulse period with IdU incorporate the second label. Due to cell renewal, autophagy and apoptosis, the amount of differentiated cells, and the amount of neoblasts that have retained label decreases with increasing chase time duration (Fig. 7A, right panel). In summary, this mode of cell turn over leads to the conclusion that random segregation of DNA-strands is the preliminary mechanism during neoblast divisions. Based on the results obtained from our label-retention study, a hypothetical graph is presented elucicating the persistance of labeled cells during the life span of M. lignano (Fig. 7B).
Figure 7. LRCs in M. lignano: their establishment, persistence and disappearance.
A): Explanatory scheme of the results of a double labeling technique, using CldU and IdU. See text for details. (B): Explanatory graph representing the curve of the number of LRCs during the lifespan of M. lignano. Two starting points correspond to hypothetical pulses, one in hatchlings and one in adults. After performing a single pulse in hatchlings (dotted line curve) or adults (full line curve), a certain number of cells incorporate the thymidine analog and become labeled. Due to proliferation and differentiation this initial population of labeled cells expands. Their progeny either retains neoblast identity (blue) or loses mitotic activity to eventually become differentiated cells (green). After an initial period of expansion, the number of CldU-labeled cells decreases as a result of cell replacement from unlabeled progeny (green curve) and dilution of the CldU in proliferating neoblasts (blue curve). After six months, a small proportion of differentiated cells and neoblasts have retained the CldU-label, respectively due to long-term functionality and cellular quiescence.
Finally, the actin-binding protein cytochalasin D was used to inhibit cytokinesis, thereby allowing analysis of the actual distribution of labeled DNA-strands among daughter cells of LRCs. This technique was performed for the first time in vivo. All binucleate cells observed, demonstrated equal distribution of labeled DNA-strands among daughter nuclei, indicating random segregation of DNA-strands in LRCs. However, it should be noted that if non-random segregation does occur, unequal distribution of fluorescence would not be visible until the second cell division after pulsing. Nonetheless, not one binucleate cell was found displaying non-random segregation of DNA-strands. Given the long chase time, and the fact that the number of LRCs was observed to decline with increasing chase time, it is unlikely that all binucleate cells divided only once after they were labeled during embryonic or post-embryonic development. In conclusion, this experiment corroborated the random distribution of DNA-strands in LRCs. The presence of a small population of quiescent neoblasts has been demonstrated previously in M. lignano. However, to date evidence was only produced for a short quiescent period of one week . In irradiation studies on M. lignano, quiescent neoblasts that were activated upon radiation, were suggested to be responsible for recovery of the animals . Our results, though, clearly demonstrate cellular quiescence on a considerably larger scale, since the foundation of a population of quiescent neoblasts appears to be already laid during the earliest stages of development. Moreover, these stem cells are shown to remain in this relatively quiescent state for a period as long as the median life-span in M. lignano.
In literature, adult stem cells are often suggested to exit the cell cycle upon reaching adulthood and form a dormant population of reserve cells. Additionally, developmental quiescence has been observed in a number of organisms. For example, in mice, the presence of quiescent or slow-cycling stem cells during the later stages of development has been observed in multiple tissues , . It is not clear however, whether these cells remain in this dormant state during adulthood. Similarly, in lower organisms, cell cycle arrest has been described for vulval precursor cells during development in C. elegans. Still, these cells already resume proliferation during a later stage of development . Thus, our observation of such an early onset of stem cell quiescence that persists for such a long time during adulthood, sheds light on a remarkable feature of neoblasts in M. lignano and opens venues for additional research.
Our study demonstrates that the neoblasts in M. lignano can be divided in at least two distinct subpopulations. The coexistence of quiescent and active neoblasts can serve to accomplish the two defining tasks of stem cell compartments, respectively maintaining a reserve for long-term repopulation, and creating progeny to cope with the high demand for proliferation. To date, it is not known whether these two populations are divided even further into a hierarchy of neoblast subpopulations with gradual limited differentiation potential. Such an organization of the stem cell pool has been postulated to greatly decrease the maximum number of cell divisions stem cells must undergo , hence reducing the risk of accumulating genomic errors. Furthermore, based on the high tolerance against radiation, neoblasts in M. lignano have been suggested to exhibit a highly efficient DNA repair system . In its natural environment M. lignano is exposed to harsh environmental conditions such as e. g desiccation, very high or low salinity and temperatures. These stress conditions can damage DNA integrity. Therefore, M. lignano might have evolved competent DNA repair mechanisms that are indirectly highly beneficial for the stem cell system.
As previously mentioned, the immortal strand hypothesis is almost impossible to reject , . Although this study produced evidence for quiescent stem cells and failed to detect non-random segregation of DNA-strands, it cannot be ruled out that only a proportion of the chromosomes are unequally distributed among daughter cells, as was reported by Armakolas and Klar . In the same way, it can never be excluded that some rare cells in the neoblasts population display non-random segregation of DNA-strands. However, the biological relevance of such a system can be questioned if it is only present in a very limited number of cells.
In this long-term in vivo study, the exact mode of DNA-segregation during stem cell division was tested in the flatworm M. lignano. Altogether, our data suggest random segregation of DNA-strands and that label-retention is a direct result of cellular quiescence. We therefore conclude that the M. lignano stem cell system is protected by the presence of a population of quiescent neoblasts, probably together with a high capacity of DNA repair. Our findings contribute to a better understanding of how stem cell systems are organized in flatworms and higher organisms, including humans.
BrdU/EdU double labeling. To test the specificity by which EdU labels cells in S-phase, an EdU/BrdU double labeling was performed. Due to unequal binding kinetics of both analogs, however, a simultaneous pulse could not be performed. Instead, hatchlings and adults were pulsed with EdU (40 min), immediately followed by a pulse with BrdU (40 min). Subsequently both EdU and BrdU were visualized. (A, B): Visualization of EdU (green) and BrdU (red) in whole mount animals (epifluorescence). Labeled cells are EdU+/BrdU+, with the exception of a small number of EdU+/BrdU− cells (arrowhead) and EdU−/BrdU+ cells (open arrowhead). These single labeled cells most likely represent cells that have left S-phase during the first pulse, and cells that have entered S-phase during the second pulse. (A): hatchling, complete animal. (B): adult, area of the gut. Abbreviations: EdU, 5-ethynyl-2′-deoxyuridine; BrdU, 5-bromo-2′-deoxyuridine. Scale Bars: 20 µm.
The authors want to thank Marjolein Couvreur for help with cultures of M. lignano, and Dr. Bernhard Egger for help with the IdU/CldU double labeling protocol.
Conceived and designed the experiments: FV MW SM PL JS. Performed the experiments: FV JS. Analyzed the data: FV MW SM KDM WB WH PL JS. Contributed reagents/materials/analysis tools: WB PL. Wrote the paper: FV PL.
- 1. Morrison SJ, Shah NM, Anderson DJ (1997) Regulatory mechanisms in stem cell biology. Cell 88: 287–298.SJ MorrisonNM ShahDJ Anderson1997Regulatory mechanisms in stem cell biology.Cell88287298
- 2. Morrison SJ, Spradling AC (2008) Stem cells and niches: Mechanisms that promote stem cell maintenance throughout life. Cell 132: 598–611.SJ MorrisonAC Spradling2008Stem cells and niches: Mechanisms that promote stem cell maintenance throughout life.Cell132598611
- 3. Sharpless NE, Depinho RA (2007) How stem cells age and why this makes us grow old. Nat Rev Mol Cell Biol 8: 703–713.NE SharplessRA Depinho2007How stem cells age and why this makes us grow old.Nat Rev Mol Cell Biol8703713
- 4. Tannenbaum E, Sherley JL, Shakhnovich EI (2005) Evolutionary dynamics of adult stem cells: Comparison of random and immortal-strand segregation mechanisms. Phys Rev E 71: 041914.E. TannenbaumJL SherleyEI Shakhnovich2005Evolutionary dynamics of adult stem cells: Comparison of random and immortal-strand segregation mechanisms.Phys Rev E71041914