S. mediterranea ETS-1 regulates the function of cathepsin-positive cells and the epidermal lineage landscape via basement membrane remodeling

The extracellular matrix (ECM) is a highly dynamic 3D network of insoluble proteins and associated polysaccharides that undergoes continuous controlled remodeling during development, tissue homeostasis and injury (Wynn, 2008; Hynes, 2009; Rozario and DeSimone, 2010; Lu et al., 2011; Xue and Jackson, 2015). In addition to the structural support for tissue integrity, ECM also facilitates receptor–ligand interaction, thereby regulating cellular processes such as proliferation, adhesion, migration, differentiation and apoptosis (Bonnans et al., 2014). ECM is also known to modulate the local concentration of growth factors and other signaling molecules by sequestering them. Therefore, tissues and cells constantly secrete factors that remodel the surrounding ECM architecture, which in turn influence their own behavior or that of the neighboring cell (Hynes, 2009). ECM is also an essential component of the adult stem cell niche and influences the ability of the niche to support stem cells and their potency (Nakayama et al., 2010; Kurtz and Oh, 2012; Mahon et al., 2021). Therefore, it is imperative to understand different mechanisms that regulate ECM architecture during development, injury and regeneration.

ECM regulators, such as the matrix metalloproteinase (MMP) family of proteases, are known to be tightly regulated at the transcriptional level. They also share common cis-elements in their promoter sequences (Yan and Boyd, 2007). Therefore, one potential mechanism of regulation for these ECM regulators is the restriction of transcription factors (TFs) in a cell-specific manner. Previous studies have identified the ETS family of transcription factors as critical regulators of extracellular matrix remodeling. The ETS transcription factor, ETS-1, has been shown to influence epithelial to mesenchymal transition (EMT) during development and in tumor invasion, by transcriptionally regulating MMPs and tissue inhibitors of metalloproteinases (TIMPs) in mammalian carcinoma cell lines (Trojanowska, 2000; Buggy et al., 2004; Rothhammer et al., 2004; Tomar et al., 2018; Zhou et al., 2018). It has also been shown that overexpression of ets-1 in human squamous carcinoma cells is associated with impaired E-cadherin localization and overexpression of MMPs, which resulted in cells showing enhanced EMT features (Taki et al., 2006).

Planarians (Schmidtea mediterranea) are an excellent invertebrate model system to study tissue turnover and regeneration due to the presence of pluripotent stem cells throughout their adulthood. However, the extrinsic factors essential for neoblast maintenance and differentiation are not well understood. The planarian homolog of ets-1 has been shown to be enriched in a functionally understudied population of cells clustered with the cathepsin-positive (hereafter denoted cathepsin⁺) population in the planarian single-cell atlas (Fincher et al., 2018). This cluster contains cell types like pigment cells, glia and dendritic cells, and has been shown to exhibit phagocytic potential (Scimone et al., 2018). A previous study has explored the role of ets-1 in pigment cell differentiation in planarians (He et al., 2017). However, the expression of ets-1 in different subclusters within the cathepsin⁺ cluster suggests that there is a broader role for this TF within a heterogeneous cathepsin⁺ population (Fincher et al., 2018).

In this study, we identified the role of the cathepsin cluster in ECM remodeling and organization. Knockdown (KD) of ets-1 led to downregulation of ECM regulators expressed in the cathepsin⁺ cluster. A detailed characterization of basement membrane revealed its thickening in KD animals. Then, we also investigated the effect of ECM thickening on neoblast migration and differentiation into the epidermal lineage. The KD of ets-1 resulted in faulty differentiation of neoblasts to late epidermal progenitors leading to a faulty epidermal turnover. In summary, our results demonstrate an evolutionarily conserved role of ETS-1 in remodeling basement membrane, thereby, controlling the epidermal lineage progression.

Single-cell transcriptome data available in the public domain (Fincher et al., 2018; GEO accession number GSE111764) was re-analyzed to study the expression of the ets-1 (dd_2092) in planarians. Our analysis revealed that ets-1 is enriched in cathepsin⁺ cell cluster (Fig. 1A; Fig. S1A,H). None of the other ETS family transcripts were enriched in the cathepsin⁺ cluster and showed mostly neuronal expression (Fig. S1I). Fluorescence in situ hybridization (FISH) showed parenchymal expression of ets-1 in the planarian body (Fig. S1B). It has been shown that irradiation of planaria leads to complete loss of neoblasts within 24–48 h post irradiation resulting in a significant reduction in the levels of transcripts that were expressed in the neoblast (Reddien et al., 2005; Hayashi et al., 2006; Palakodeti et al., 2008). To test the expression of ets-1 in neoblast versus differentiated tissue, ets-1 transcript levels were measured post lethal irradiation, which showed a 30% reduction at 72 h (Fig. S1D). This suggests that the majority of ets-1 expression is in differentiated tissue rather than the neoblast. Furthermore, we also compared the expression of ets-1 in the piwi-1 compartment, which marks the neoblast population. Within the piwi-1⁺ cluster, the expression of ets-1 is enriched in the cathepsin⁺ subcluster (Fig. 1A,C; Fig. S1A) suggesting that ets-1 might play an important role in the differentiation of a subset of neoblasts towards the cathepsin⁺ lineage. We validated ets-1 expression in the cathepsin⁺ cells by performing double FISH to study the co-expression of ets-1 with foxf-1, a known regulator of cathepsin⁺ cells (Scimone et al., 2018). These studies revealed high co-expression, foxf-1⁺ cells (73%) with ets-1⁺. Furthermore, ets-1 also showed co-expression with PIWI-1, which marks the progenitor population (Fig. S1C). Co-expression of foxf-1⁺ and ets-1⁺ with PIWI-1 showed that 30% of foxf-1 and ets-1 double-positive cells were PIWI-1⁺, suggesting this population could potentially be the cathepsin⁺ progenitor pool (Fig. 1A; Fig. S1C).

To decipher the function of ETS-1, RNAi was performed on intact planarians by feeding them ets-1 double-stranded (ds)RNA by following a feeding regime as depicted (Fig. S1E). The intact animal treated with ets-1 dsRNA developed of anterior lesions, followed by the lysis of the anterior half (Fig. 1C; Fig. S1F) and eventual death in ∼20–30 days from the first feed (Fig. 1B). For better characterization of the phenotype, transcriptome sequencing was performed from the ets-1 KD animals at 7 days post last feed, before the appearance of head lesion phenotype. The gene expression analysis showed 291 genes (adjusted P<0.05) that were differentially regulated in the ets-1 KD animals, of which 224 transcripts were downregulated. The majority of the downregulated genes were expressed in the cathepsin⁺ cluster (173/224 genes; Fig. 1D; Table S1) suggesting that ets-1 is essential for the function of the cathepsin⁺ cell cluster. Our transcriptome data from the ets-1 KD also showed no significant decrease in the expression of the genes encoding other ETS family of proteins, validating the specificity of the ets-1 RNAi (Fig. S1G). The cathepsin⁺ cell cluster contains various cell types such as glia, pigment cells and other dendritic cells (Fincher et al., 2018; Scimone et al., 2018). Transcriptome analysis of ets-1 KD animals showed significant downregulation (P<0.05) of various cathepsin⁺ cell subcluster markers, for example, dd_10181, aqp-1, CTSL-2, dd_7593 and cali, in addition to fox-f1, a known marker for a majority of the cathepsin⁺ cluster. This suggests that ets-1 RNAi affects multiple cathepsin⁺ subclusters (Fig. 1E). Transcriptome sequencing showed downregulation of cathepsin⁺ cluster transcripts in ets-1 KD animals, as validated by performing FISH using the markers for various cathepsin cell types (cali for glia, ctsl-1 and apq-1 for dendritic cells). We observed a reduction in the number of cells expressing these transcripts in ets-1 KD worms (Fig. 1F) validating the transcriptome data. Furthermore, the expression of foxf-1 was also downregulated in FISH and quantitative (q)PCR data in ets-1 KD animals (Fig. 1G,H). Taken together, our results suggest that ets-1 is critical for maintenance of foxf-1⁺ cathepsin cluster.

Recent computational analysis has shown that core structural ECM proteins, such as collagens and other glycoproteins are expressed in the muscle cell type (Cote et al., 2019). Our detailed analysis of planarian matrisome (ECM and ECM-associated proteins) genes showed expression of a subset of ECM regulators in the ets-1-expressing cathepsin⁺ cluster (Fig. S1H). In ets-1 KD animals, the expression of matrisome genes specific to cathepsin cluster was significantly downregulated (P<0.05, 14/35 genes; Fig. 2A; Table S1). The majority of these downregulated genes were ECM regulators (11/14 genes; Fig. S2A). Transcriptome data showing the downregulation of ECM regulators, such as dd_2305, dd_198 and dd_2307, was verified by FISH. All three genes showed downregulation in the ets-1 RNAi animals compared to their controls (Fig. S2B). In addition, other ECM modulators and four core glycoproteins that were expressed in cathepsin⁺ cluster were downregulated in the ets-1 RNAi as shown by qPCR (Fig. S2C,D). Given that ECM regulators are downregulated upon ets-1 RNAi, we hypothesized that the defect in ECM remodeling leads to the accumulation of ECM proteins resulting in the thickening and disorganization of the basement membrane. Histological section stained using Hematoxylin and Eosin (H&E) showed that there was an increase in basement membrane thickness in ets-1 KD. We further validated this observation by transmission electron microscopy (TEM), which showed an increase in basement membrane thickness in the ets-1 RNAi animals (Fig. 2B). Given that collagen-IV is the most abundant core structural protein of ECM, we stained for collagen-IV on the transverse section of ets-1 KD and control animals. Transverse sections from ets-1 RNAi animals also showed increased collagen-IV in the basement membrane and ruffling of basement membrane edges (Fig. 2B,C). We also observed an overall increase in collagen-IV intensity lining muscle fibers in the planarian body (Fig. 2C,H). We quantified the thickness of the basement membrane along the dorsal and the ventral side in ets-1 KD animals. A significant increase (P<0.05) in thickness of basement membrane (measured using collagen-IV) was observed both at the dorsal and ventral side in ets-1 RNAi animals (Fig. 2D–F). Transcriptional upregulation of collagen-IV as a reason for increased thickness was ruled out from FISH of collagen-4 and the transcriptome data from ets-1 KD animals, which did not show any significant increase in the levels of collagen-4 transcripts (Fig. S2E–G). Together, our data shows that the increased collagen-IV is potentially due to the absence of ECM regulators in the KD animals.

Remodeling of tissue is an indispensable part of successful regeneration. Thus, we looked at basement membrane organization in ets-1 RNAi. We used sagittal regeneration as our proxy to compare regenerating and non-regenerating tissue at the same position across the anterior–posterior (A-P) axis. The four feed RNAi paradigms used to study the homeostatic phenotype in ets-1 RNAi animals (Fig. 1) led to lysis in regenerating animals in 5 days post amputation (Fig. S3B). Therefore, to study the regeneration defect, we have used a three-feed paradigm separated by 3 days followed by amputation at the third day post last feed (Fig. 3B). We observed 60–70% animals survived the regeneration paradigm until 15 days post amputation (dpA) (Fig. 3A,B). However, the body size of the ets-1 KD animals was reduced by 20–30% (Fig. 3C) and the animals shows varied levels of regeneration defects, such as lysis (n=11/40) or smaller blastema (n=16/40). We selected animals with smaller blastema to access the regeneration defects. Immunostaining for photo receptor neurons (VC-1) and FISH for an anterior gut branching marker (Mat) shows that there are defects in the organization of eye and gut in ets-1 KD animals. Despite having small blastema we did not notice any decrease in the number of piwi-1⁺ cells and collagen⁺ muscle cells compared to the control animals (Fig. S3A). We performed RNA sequencing in ets-1 KD regenerating animals at 5 dpA. We observed a reduction in expression of cathepsin⁺ cluster transcripts similar to that seen with ets-1 RNAi during homeostasis (Fig. 3D,E; Table S1). We also found the expression of 12 ECM regulators and 10 core glycoproteins was significantly downregulated (P<0.05) (Fig. 3F,G). Further, we also observed increase in collagen-IV thickness throughout the body both on the regenerating and non-regenerating side (Fig. 3H–J). Together, our results show that ETS-1 is a key transcription factor which controls the expression of ECM regulators essential for maintaining ECM organization during regeneration.

Our study shows the important role of ETS-1 in the regulation of ECM modulators. A similar regulation of ECM by ETS family of proteins has also been shown in several tumors (Buggy et al., 2004; Ghosh et al., 2012). Therefore, we reasoned that the underlying cause for the collagen-IV thickness in ets-1 KD animals could be downregulation of ECM modulators in the cathepsin⁺ clusters. To test this hypothesis, RNAi was performed on five ECM modulator genes selected from the transcriptome data that were downregulated after ets-1 RNAi [tolloid like-1 (dd_1967), tolloid like-2 (dd_2307, dd_3177), chymotrypsin-like elastase (dd_198) and mt-mmpA (dd_2305); Fig. 4A, Fig. S4A]. Knockdown of mt-mmpA, a known target of ETS-1, gave rise to lesions on the anterior side, similar to what was seen with ets-1 RNAi animals, and so was selected for further phenotype characterization. Single-cell transcriptome analysis for the expression of mt-mmpA suggested its expression in cathepsin⁺ cluster, similar to ets-1 (Fig. S4B,C). FISH for mt-mmpA revealed its broad expression across the whole body, mostly in the cells right below the basement membrane and that spread throughout the parenchyma except in the gut and pharynx (Fig. 4B). Furthermore, double FISH revealed co-expression of ets-1 and mt-mmpA (Fig. 4C), confirming that mt-mmpA is expressed in the same compartment as ets-1. We undertook mt-mmpA FISH along with collagen-IV immunostaining, which showed the expression of mt-mmpA in the cell types just below the basement membrane (Fig. 4D). A reduction in mt-mmpA expression was observed upon ets-1 RNAi (Fig. 4E), which was validated by ets-1 KD transcriptome data and qPCR data (Figs S4D and S2C), suggesting that mt-mmpA is downstream of ets-1. Given that ets-1 KD leads to thickening of the ECM, we also investigated the basement membrane thickness on the transverse sections from mt-mmpA RNAi animals. Transverse sections from the mt-mmpA KD animals showed a significant increase (P<0.05) in collagen-IV thickness (Fig. 4F,G), which recapitulates the phenotype observed in the ets-1 KD animals. Together, our data suggests that ETS-1 is essential for the maintenance of cathepsin⁺ cluster, which expresses ECM modulators, such as mt-mmpA, and that ets-1 KD leads to a defect in the maintenance of the basement membrane thickness and ECM organization (Fig. 4H).

In planarians, reduction in the ECM thickness has been shown to affect the progression of neoblasts and their differentiation (Cote et al., 2019; Lindsay-Mosher et al., 2020). Our study showed that the knockdown of ets-1 caused thickening of ECM, which led us to investigate the role of ets-1 KD on neoblast proliferation and differentiation to specific lineages. First, we investigated for the change in the total number of cells in the neoblast population upon ets-1 KD by performing FISH for smedwi-1, a known neoblast marker. There was no significant change in the total number of neoblasts in ets-1 KD animals compared to that in the control (Fig. S5A–C). The proliferation state of the neoblast was also measured by counting the number of cells positive for phosphorylated histone 3 (H3P⁺) cells, which marks neoblasts in the G2/M phase of the cell cycle. Immunostaining using an antibody against H3P showed no significant change in number of H3P⁺ cells in the ets-1 KD animals suggesting the ETS-1 is not critical for either maintenance or proliferation of the neoblast (Fig. S5D,E). Previous work (Cote et al., 2019; Lindsay-Mosher et al., 2020; Chan et al., 2021) has shown that the downregulation of ECM-associated genes resulted in the mis-positioning of neoblasts towards the lateral edge of the animal. We hypothesized that RNAi of ets-1, which led to thickening of ECM, might result in the abnormal change in the spatial compartmentalization of neoblasts. To test this, we measured the distance of piwi-1⁺ cells from the edge of the epidermis in control and ets-1 KD animals. The ets-1 KD animals showed a significant increase (P<0.05) in the distance between the piwi-1⁺ cells and the lateral edge of the animals (Fig. 5A–C). Altogether, these results suggest that the piwi-1⁺ neoblast compartment is restricted towards the medial plane in the KD animals compared to the controls.

It has been shown that the perturbation of mt-mmpA leads to a defect in the migration of the neoblast population (Isolani et al., 2013). Since ets-1 RNAi leads to mt-mmpA downregulation, we investigated whether there was a neoblast migration defect in the ets-1 KD animals. For the migration studies, a BrdU chase experiment was performed by feeding BrdU to control and ets-1 KD animals. To rule out that there was a defect in global differentiation, we measured the number of piwi-1⁺ and BrdU⁺ cells at 1 day and 4 day post BrdU in ets-1 RNAi. The percentage of piwi-1 and BrdU double-positive cells in ets-1 KD animals were comparable to control at both time points suggesting that there was no global differentiation defect (Fig. S5F,G). To access the migration defect, the number of BrdU⁺ cells at the anterior tip of the animal right above the eye was quantified. Typically, the anterior tip lacks neoblasts. The presence of the BrdU⁺ cell at the tip is an indication that neoblasts have differentiated into progenitors and migrated towards the tip. A decreased number of BrdU⁺ cells at the anterior tip is an indication of defective migration for the neoblast and its progenitors. In the ets-1 KD, we observed a significant decrease (P<0.05) in the number of BrdU⁺ cells at 3 day BrdU chase (Fig. 5D,E), which suggests that regulation of ECM modulators by ets-1 controls the organization of the ECM, which is critical for the migration of the neoblast and progenitors.

Earlier published work has shown that the early and late progenitor populations are spatially segregated in planarians (Tu et al., 2015; Zhu and Pearson, 2018). As the neoblast differentiates, the early and late progenitors tend to migrate away from the medial plane. Based on our observations that ets-1 KD leads to thickening of the ECM, restriction in the spatial positioning of the neoblast and subsequent defects in migration, we hypothesize that the KD animals might show defect in differentiation of neoblasts and mis-positioning of the progenitor population. Of the many cell lineages in planaria, epidermal lineage progression has been well defined, with distinct spatial and temporal markers identified for both early and late epidermal progenitors (Tu et al., 2015). Therefore, epidermal lineage progression serves as a good model to study the neoblast differentiation in ets-1 KD animals. We investigated the epidermal lineages using markers for early progeny (prog-1), late progeny (AGAT-1), later progeny (Zpuf-6) and terminal epidermis (vim-1). FISH followed by quantitative analysis showed a reduction in the number of early progeny cells (prog-1) in the ets-1 KD animals (Fig. 5F). Although we noticed a reduction in the prog-1⁺ cells, the late progeny (AGAT-1⁺) cell numbers were comparable to those in the control (Fig. 5F–H). This is also confirmed by qPCR data, which also showed a significant decrease (P<0.05) in prog-1 and zfp-1 transcripts in ets-1 KD animals (Fig. S5H). We also noticed a decrease in the number of terminal differentiated epidermal cells marked by vim-1 both by FISH and qPCR in ets-1 KD animals. Earlier work has shown that early and late progenitors are at a distinct spatial position in the animal with early being closer to the medial plane compared to the late progenitors, which are towards the lateral edge (Tu et al., 2015). In control animals, AGAT-1⁺ cells were organized just below the basement membrane in a single array of cells. However, in the ets-1 RNAi animals, the AGAT-1⁺ cells were found to be distributed in multiple layers inside the planarian body away from the epidermis, which was quantified by measuring the distance of AGAT-1⁺ cells from the epidermal layer (Fig. 5I,J). PIWI-1 protein in the planarian is expressed both in the neoblasts and the early progenitors, whereas the piwi-1 mRNA is exclusively present in the neoblasts (Reddien et al., 2005). Given that the late epidermal progenitor cells were distributed in the early progeny compartment in the ets-1 KD animals, we investigated the presence of PIWI protein in AGAT-1⁺ cells. Interestingly, a 3-fold increase in the number of AGAT-1⁺ cells expressing PIWI-1 protein was observed, suggesting an accelerated differentiation of PIWI-1⁺ cells to AGAT-1⁺ late progeny (Fig. 5K,L). Furthermore, we also performed a 2-day BrdU chase experiment in ets-1 KD animals to show the accelerated differentiation of neoblasts to late progenitors. It showed a 2-fold increase in the number of BrdU⁺ cells expressing AGAT-1 in the ets-1 KD animals (Fig. 5M,N). Together, our data suggest that the thickening of ECM in the ets-1 RNAi animal could potentially lead to a spatial and temporal shift in the differentiation of neoblasts, which is evident from the epidermal lineage progression. However, it is important to verify whether such perturbation in differentiation upon ets-1 KD is seen in other lineages or is specific to the epidermal lineages.

ECM, which is the main component of the micro-environment of stem cells niche plays a pivotal role in regulating stem cell maintenance and differentiation. However, the mechanism that controls ECM organization at the organismal level essential for regeneration of tissues is not well understood. It has been shown that the ECM undergoes extensive remodeling to maintain its thickness and organization, which is primarily driven by the ECM regulators, such as metalloproteinases (Page-McCaw et al., 2007; Lu et al., 2011). Most of our current understanding of the role of ECM and its regulators is driven by studies on tumor progression and metastasis (Winkler et al., 2020). The ECM regulators are under strict transcriptional regulation. For example, one of the most common cis-elements found in promoters of ECM regulators (MMPs) is the PEA3 site, which can bind to multiple ETS factors (Trojanowska, 2000; Chakraborti et al., 2003). Most of the functional studies associated with ETS in ECM regulation are restricted to tumorigenesis (Kar and Gutierrez-Hartmann, 2013; Zhou et al., 2018; Hsing et al., 2020). A constitutive ETS-1-knockout model also shows a defect in immune cell differentiation, particularly in the specification of natural killer cells, dendritic cells and T-lymphocyte maturation (Bartel et al., 2000). However, its role in regulation of stem cell proliferation and differentiation during development and regeneration is not well understood.

In this study, using planaria, we demonstrated the role of ETS-1 in regulating the thickness and organization of ECM essential for stem cell migration and differentiation. Previous studies on ECM in planaria were focused on ECM structural protein. For example, knockdown of collagen IV, hemicentin and megf6 has been shown to disrupt the basement membrane, thereby affecting the spatial distribution and migration of neoblasts as well as fate commitment (Cote et al., 2019; Lindsay-Mosher et al., 2020; Chan et al., 2021). Our current study identified the conserved role of ETS family of transcription factor in regulating ECM regulators critical for planarian homeostasis and regeneration.

The planarian homolog of ets-1 has been shown to affect the differentiation of the pigment cell lineage, which is a subset of cathepsin⁺ cell population. In particular, it has been shown that knockdown of ets-1 results in downregulation of punctate cell marker (LysoLP-1) leading to de-pigmentation (He et al., 2017). Our study identified the expression and function of ets-1 in cathepsin⁺ cell types, a tissue type whose function has not been well characterized. Single-cell RNA sequencing also confirmed the expression of ets-1 as well as a majority of other ECM modulators within this cell type (Fincher et al., 2018; Cote et al., 2019). In ets-1 RNAi animals, downregulation of ECM regulators such as tolloid like-1 (dd_1967), tolloid like-2 (dd_2307 and dd_3177), chymotrypsin-like elastase (dd_198) and mt-mmpA (dd_2305) suggests that ets-1 controls the expression of ECM regulators within the cathepsin⁺ cell cluster. Similar regulation of MMPs by ETS1 has been shown in mouse stratified epithelium (Nagarajan et al., 2010) and multiple other cancer studies (Behrens et al., 2001; Ghosh et al., 2012; Nazir et al., 2019).

In ets-1 RNAi, we observed a basement membrane defect underlying the epidermal layer as evidenced by the increase in the collagen-IV protein, which is the primary component of basement membrane (Frantz et al., 2010). Our FISH and qPCR data to measure collagen-IV transcript levels ruled out the transcriptional upregulation of collagen-IV in the ets-1 KD animals. This suggests that downregulation of ECM regulators as the primary cause for the increase in the collagen-IV levels. Knockdown of mt-mmpA, which is a major ECM regulator, phenocopied ets-1 KD. Human ChIP seq data also shows that MMP24, the human homolog of MT-MMPA, is a direct transcriptional target of ETS1 along with multiple other ECM regulators (Rouillard et al., 2016). Furthermore, in planarians, mt-mmpA RNAi was shown to affect the migration of neoblast progeny, which was also observed in ets-1 KD animals. These results suggest that mt-mmpA is a likely downstream target of ETS-1 in planaria, and is critical for basement membrane organization. Our study also highlights an interesting paradigm in planaria, in which most of the structural ECM proteins are expressed in muscle and that ECM regulators, such as mt-mmpA, are enriched in cathepsin⁺ cells. This suggests that a communication between these two cell types, cathepsin⁺ and muscle, is critical for the maintenance of ECM thickness. Here, we put forth two possibilities that could explain the potential crosstalk between the two cell types: (1) that proteases, like MT-MMPA, can be tethered to the outer leaflet of the plasma membrane of cathepsin⁺ cells for collagenase activity as shown by Fillmore in glioma invasion (Fillmore et al., 2001); and (2) that MT-MMPs can activate other MMPs, which are then secreted in the extracellular space on appropriate signaling thereby breaking down the structural ECM proteins like collagen, perlecan, etc. (Fillmore et al., 2001; Itoh, 2015). Therefore, our results suggest a potential crosstalk between muscle cells and the cathepsin⁺ cells (Fig. 6A) for the maintenance of basement membrane. However, the regulatory mechanism that modulates the cross between these two cell types critical for the organization of the ECM needs further investigation.

In planarians, knockdown of ECM proteins has been shown to affect the boundary of the stem cell compartment (Cote et al., 2019; Lindsay-Mosher et al., 2020). Our ets-1 KD studies have shown no change in neoblast maintenance and proliferation. However, comprehensive analysis of various epidermal progenitor markers, as a proxy to understand the effect of ECM thickening on neoblast lineage commitment, showed disrupted stem cell compartment boundaries and faulty differentiation of epidermal progenitor. Our results show that the epidermal early progeny, marked by prog-1, showed a marked reduction in their numbers, whereas, late epidermal progeny (AGAT-1, Zpuf-6) do not show drastic reduction. Surprisingly, ets-1 KD animals show an increased number (∼3-fold higher) of late progenitor cells, marked by AGAT-1 and co-expressing PIWI-1, which under normal conditions is restricted to neoblasts and their early progenitors. Furthermore, a BrdU chase experiment revealed accelerated progression of BrdU⁺ cells to late epidermal progeny in the ets-1 KD animals. At this point, it must be noted that ets-1 was not enriched within the epidermal lineage in single-cell RNA sequencing datasets, suggesting that the defect in epidermal lineage progression is not because of the direct regulation of epidermal lineage markers by ets-1 (Fincher et al., 2018). Therefore, together, these results show that a thickened basal lamina perturbs the bona fide lineage progression trajectory in a stem cell non-autonomous manner. However, what happens to the lineage progression of other tissue types, such as muscles or neurons, is not known at this time point and remains to be investigated. The cues that are critical for the neoblast differentiation could either be mechanical, driven by the stiffness of the ECM, or be mediated through trapping of morphogens. Thus, the planarian stem cell model provides an ideal platform to dissect the ECM-mediated mechanisms that could potentially regulate spatial and temporal differentiation of neoblasts to specific progenitors.

Finally, our data also shows that ets-1 KD results in the downregulation of mesenchymal markers such as CTSL-2, APQ-1 and CALI-1, which are associated with cathepsin⁺ clusters. These markers are associated with cell population(s) that are broadly clustered under the cathepsin⁺ population in the planarian single-cell atlas due to high expression of cathepsin proteins (lysosomal enzymes/proteases) in their transcriptome (Fincher et al., 2018). This cluster includes cell types such as pigment cells, glia-like cells, and other dendritic and phagocytic cell types. However, the function of these cell types in regulating different aspects of planarian biology is poorly understood. Our understanding about the regulation of these cell types is currently limited to the TF foxf-1 (Scimone et al., 2018). It has been shown that in the absence of foxf-1, the specification and/or maintenance of multiple cathepsin⁺ populations, such as pigment cells, glia-like cells and phagocytic cells, is affected. Our study shows a similar trend. In addition, knockdown of ets-1 results in decreased expression of foxf-1, suggesting ETS-1 is an upstream regulator of FOXF-1, which is critical for cathepsin⁺ cluster maintenance. The potential role of ETS-1 in regulating forkhead TFs is also supported by a human ChIP-seq dataset, which suggests forkhead TFs are one of the targets of ETS transcription factors (Rouillard et al., 2016). Several studies from vertebrate and invertebrate model systems show that the ETS family of proteins is a key regulator of dendritic and phagocytic cells; the ETS-1 homolog is required for the specification of radial glial cells, astrocytes, dendritic cells and natural killer cells in mammals (Amouyel et al., 1988; Maroulakou and Bowe, 2000; Kiyota et al., 2007), suggesting a functionally conserved role of ETS-1 in planarians.

Therefore, to summarize, in addition to the phagocytic function of cathepsin⁺ cells, our study ascribes a new function of this cell cluster in regulating ECM thickness, which in turn is critical for stem cell fate and function.

The Schmidtea mediterranea sexual stain was used in all the experiments in the study. Stock worms were maintained in 5 l plastic boxes containing 1× planaria medium [composition: 1.6 mM NaCl, 1 mM MgSO₄, 0.1 mM MgCl₂, 0.1 mM KCl, 1 mM CaCl₂ and 1.2 mM NaHCO₃ (Cebrià and Newmark, 2005)] in Milli-Q water. The planaria medium was filtered using a vacuum filter system with 0.45 µm filters (Membrane filters, Millipore_654). Lids were loosely placed over the boxes and stored in an incubator (SANYO MIR-554) in the dark at 18–20°C. The worms were fed beef liver paste twice a week for the colony maintenance. Size-matched animals were starved for at least 7 days before using in any experiment or before starting RNAi feeding. For RNAi experiments, worms were kept in standard Petri dishes (120 mm) and cleaned every other day.

For extracting RNA from planaria, RNAiso Plus (total RNA extraction reagent cat: #9109) was used. Briefly, the worms/tissue were collected in at least 300 µl of RNAiso and stored at −80°C for 1 h/overnight. The tissue was homogenized using a pestle in a 1.5 ml Eppendorf tubes. A 1/5th volume of chloroform was added and mixed by inverting, and kept on ice for 10 min with occasional inverting. Samples were centrifuged at 21,000 g for 25 min at 4°C. The aqueous phase was collected carefully, without disturbing the interphase, and placed into a fresh tube. An equal volume of pre-chilled isopropanol was added, mixed and kept at −20°C for 30 min. RNA pelleting was undertaken by centrifuging at 18,000 g for 20 min. A dark brown pellet of RNA was observed and washed twice with 70% ethanol (in nuclease-free water) by centrifuging at 18,000 g for 10 min at 4°C. The pellet was air dried and re-suspended in nuclease-free water; RNA was always stored at −80°C.

1–2 µg of RNA was used to make cDNA using SuperScript™ III Reverese Transcriptase (Invitrogen cat. no. 18080044) using oligo dT₂₀ primers (Invitrogen cat. no. 18418020). The reaction was set up according to the protocol manual using RNaseOUT™ Recombinant (Invitrogen cat. no. 10777019). RNA complementary to the cDNA, was removed by treating with RNase-H (Invitrogen cat. no. 18021-014) at 37°C for 20 min.

The gene-specific primer was used to amplify the gene using Takara LA Taq DNA Polymerase (cat. no. RR002C) and cloned using a TA cloning kit (Invitrogen cat. no. K207040) as per the protocol manual. The ligated construct was transformed into competent Escherichia coli DH5α cells, and positive clones were picked using blue–white colony screening in a kanamycin resistance background. The clone was validated and the orientation of the gene was determined by Sanger sequencing using M-13 primers (Invitrogen cat. nos. N52002 and N53002).

Either a cloned plasmid was used as a template for dsRNA for probe synthesis using T7 polymerase (Invitrogen cat. no. AM2718) and SP6 polymerase (Roche cat. no. 10810274001), or a T7 promoter overhang-gene-specific primer amplified template was used for T7 polymerase-dependent in vitro transcription to produce dsRNA. For probe synthesis, Dig RNA labeling mix (Roche cat. no. 10810274001) or Fluorescein RNA labeling mix (Roche cat. no. 11685619910) was used in an in vitro transcription reaction.

Quantitative real-time PCR (qPCR) was performed using the PowerUP SYBR Green master mix (Thermo Fisher Scientific; cat. no. 4367659) on a Quant Studio 5 system (Applied Biosystems). The data was downloaded and analyzed in Microsoft Excel. Primer sequences used in the study can be found in the Table S2. All samples were run in triplicates and each experiment was repeated two times. Gene expression was determined by the 2^−ΔΔCt method and the actin transcript expression levels were used as internal control to normalize the expression level for different samples.

Purified dsRNA was mixed with beef liver extract in 2 µg/10 µl (dsRNA/liver extract). The aliquots were stored at −20°C before use. dsRNA against green fluorescent protein (GFP), which is not expressed in planaria was used as RNAi control for comparisons. The feeding strategy used for RNAi during homeostasis conditions was four dsRNA feedings with a 3-day gap followed by a 7-day wait before fixing (Rouhana et al., 2013).

For regeneration, the animal was fed three times with dsRNA with a 3-day gap followed by a 2-day wait before amputation.

For the RNAi screen for downregulation of ECM regulators, 12 feeds were given spaced by a 5-day gap.

A 20 mg/ml stock solution of BrdU (Roche cat. no. 10280879001) in 50% ethanol dissolved in 50% DMSO in 1× Planaria medium feed was used, with 20 µl of the stock solution in 80 µl of beef liver extract.

Animals used in BrdU chase experiments were adapted to high salt medium (5× planaria medium) for 3 days prior to BrdU feeding (Newmark and Sánchez Alvarado, 2000; Sarkar et al., 2022). The day of feeding was considered day 0 and the chase was for the desired duration. The animals were killed using 5% N-acetyl cystine (Sigma cat. no. A7250) in 1× PBS for 7 min, then washed twice with 1× PBS followed by fixation in pre-chilled Meth–Carnoy solution (6:3:1 methanol:chloroform:acetic acid by volume) for 40 min at 4°C with mild rocking. After five washes with 100% methanol animals were bleached using 6% H₂O₂ in 100% methanol. Animals were washed three times with 1× PBS with 0.3% Triton X-100 [PBSTx_(0.3%)] for 10 min each before incubation in 2 M with 0.5 Triton X-100 [HCl-Tx_(0.5%)] for 40–50 min at room temperature followed by a 5-min incubation in 0.1 M sodium borate. Animals were washed again three times with 1× PBSTx_(0.3%) for 10 min each before being blocked in blocking solution (5% Roche western blocking reagent and 5% horse serum in 1× PBS) for 2 h at room temperature. Anti-BrdU antibody (Abcam cat. no. ab6326) was diluted (1:1000) in blocking solution and animals were incubated in it at 4°C for 40 h. Animals were washed again three times with 1× PBSTx_(0.3%) for 10 min each. Animals were incubated with Alexa Fluor-conjugated secondary antibodies (dilution 1:1000) and DAPI (Sigma Aldrich cat. no. D9542, dilution 1:1000) for 1 h at room temperature. Animals were washed again three times with 1× PBSTx_(0.3%) before mounting on glass slides using Fluoroshield (Sigma-Aldrich cat. no. F6182).

For vibratome sectioning, a LEICA VT 1200S was used. Samples were fixed in Carnoy's fixative (6:3:1 ethanol:chloroform:acetic acid by volume). Animals were rehydrated by passing through an alcohol gradient (100%, 70%, 50% and 25% ethanol, then 1× PBS) and embedded in 4% low-melting agarose (Invitrogen cat. no. 16520050) with desired orientation. The block was trimmed and stuck on the cutting stage using Fevi kwik (industrial glue). The collecting chamber was filled with pre-chilled PBS and 40 µm sections were collected in cryoprotectant solution (PBS:glycerol:ethylene glycol, 4:3:3), and stored in −20°C for at least overnight before staining.

For microtome sectioning a LECIA RM2235s was used for the sectioning (10 µm section). Samples were fixed in Carnoy's fixative, washed with ethanol (five times), and clearing of the samples was done with xylene for 10 min and infiltrated with paraffin wax [first, using paraffin:xylene (1:1) for 10 min in an oven pre-set at 60°C and then transferring to melted paraffin wax for a 30 min incubation]. The paraffin wax was changed twice before the block was taken out and placed on ice for solidification. These blocks were stored at room temperature until sectioning.

The primers used for making riboprobes can be found in Table S2. RNA-FISH for ets-1 was performed using a cocktail of three non-overlapping probes with a concentration of 1 ng/µl each. For FISH, worms were killed using 5% N-acetyl cystine for 5 min and fixed in 4% formaldehyde (in 1× PBS) for 20 min and dehydrated in gradient of ethanol and stored at −20°C. The samples were rehydrated and bleached with 5% formamide (with 1.2% H₂O₂) in 1× PBS under bright light. Then samples were permeabilized for riboprobe access using 10 min proteinase-K treatment followed by 4% formaldehyde fixation. Riboprobes were prepared using digoxigenin (DIG) or fluorescein (FITC) labeling mix. The probe was diluted in reducing hybridization buffer and incubated for 16 h in a hybridization oven with shaking at 56°C. The samples were washed with TNTx buffer [0.1M of Tris-HCl (pH 7.5), 0.15M NaCl, 0.05 % Triton X-100] and incubated in blocking for 2 h at room temperature (RT) with gentle rocking in 5% horse serum and 5% Roche western blocking reagent in TNTx. Followed by overnight incubating in anti-Dig POD (1:1000) or anti-fluorescein POD (1:1000) at 4°C. The signal was developed using the tyramide signal amplification method (King and Newmark, 2013).

The samples were fixed in either Carnoy's solution or 4% formaldehyde solution in PBSTx_(0.3%). Samples were bleached using 6% H₂O₂ under bright light overnight. They were washed three times with methanol at RT followed by rehydration using 50% methanol, 25% methanol and 1× PBS with 0.55% Triton X-100 for 10 min and then processed according to a previously described protocol (Sarkar et al., 2022). The following primary antibodies were used in the study: anti-collagen-IV (Abcam cat. no. ab6586, 1:200), anti-VC1 (a kind gift from Kiyokazu Agata, National Institute for Basic Biology, Okazaki, Japan; 1:2000), anti-BrdU (Abcam cat. no. ab8152, 1:1000), anti-H3P (Invitrogen cat. no. PA5-17869, 1:200), anti-PIWI-1 (1:200; Guo et al., 2006), and anti-SMED 6G10 (DHSB cat. no. 6G10 2C7, 1:500; Ross et al., 2015). Nuclei were stained with DAPI (1:1000 from 5 mg/ml stock; Sigma-Aldrich cat. no. D9542).

For H&E staining, 10 µm thick paraffin sections were used. Sections were rehydrated using an ethanol gradient (100%, 75%, 50%, 25% and 0% in Milli-Q water) and immersed in Mayer's Hematoxylin and agitated for 30 s and rinsed immediately in Milli-Q water. Slides were immersed in 1% Eosin solution for 10–30 s with agitation. The sections were dehydrated with two washes of 95% ethanol and 100% ethanol for 1 min each. Samples were immersed in xylene twice before mounting using DPX. The protocol was adapted from Fischer et al. (2008).

An Olympus SZX16 or Olympus IX73 microscope was used to take all the brightfield/darkfield images. The primary microscope used to acquire confocal image was a FV3000 laser scanning microscope using 20×0.8 NA or 40X 1.3 NA objective (Olympus) with Olympus Flow view software. The images were processed and quantified in Fiji (open-source software).

For transmission electron microscopy, sample preparation was performed as previously described (Brubacher et al., 2014). The RMC Powertome ultramicrotome was used to section the resin block. For image acquisition, a MERLIN Compact VP was used in STEM mode.

Basement membrane analysis in homeostasis was performed using 10 free-floating vibratome sections per worm for ets-1 RNAi and mt-mmpA RNAi. For regeneration, four free-floating vibratome sections per worm were stained with anti-collagen-IV antibody along with anti-6G10 (a planarian muscle marker) and DAPI (nuclear dye). Three regions of interest (ROIs) were imaged with a 40× oil objective from dorsal and ventral surface of the slices. Using Fiji, the width of the basement membrane marked by collagen-IV was recorded manually at five different locations in an ROI using line tool and averaged per section.

Using Fiji, the number of cells co-stained with each marker were counted manually in the ROI. The number of foxf-1 and PIWI-1⁺ cells also co-expressing ets-1 was divided by total number of ets-1⁺ cells to calculate the percentage of double-positive cells.

Using Fiji, the number of AGAT-1 positive cells in the entire the Z-stack was counted for the ROI shown in the schematic, and expression of PIWI-1 was analyzed in these cells to report the percentage of double-positive cells.

Using Fiji, the number of BrdU⁺ cells in the entire Z-stack was counted at 2 days post BrdU addition, and expression of AGAT-1 was analyzed in these cells to determine the percentage of double-positive cells.

For analysis of the cell distance from lateral edges, the distance of the piwi-1⁺ neoblast or AGAT-1⁺ cells from the epidermal edge was manually calculated throughout the Z-stack in Fiji software. The distance was plotted in the graph.

For analysis of cell migration, BrdU⁺ cells above the photoreceptors were manually counted throughout all the Z-stack and normalized to the area of the animal above the photoreceptors in Fiji software. The graph was plotted as BrdU⁺ cells/100 mm² area.

For analysis of cell number, the ROI shown in the schematic was imaged and cell counting was undertaken throughout the Z-stack. The cells positive for the cell marker were counted and normalized to the area of the ROI. The graph was plotted as cell marker-positive cells/100 mm² area.

For NB.22.1e the quantification was done from the lateral edge of epidermis and the graph was plotted as NB.22.1e⁺ cells/100 µm of the epidermal edge.

RNA was extracted from the whole animal (at day 7 post last feed or, for regenerating fragments, at 5 days post sagittal amputation). A transcriptome library was prepared using NEBNext^® Ultra™ II Directional RNA Library Prep with Sample Purification Beads (cat. no. E7765L) kit and sequenced in a Illumina HiSeq 2500 machine. All the samples were sequenced (as single end) in biological replicates (for homeostatic condition in duplicates and sagittal regeneration in triplicate). Approximately 20–25 million reads were sequenced for every sample. These reads were adapter trimmed using Trimmomatic (Bolger et al., 2014) and mapped to rRNA and other databases to remove contamination. Reads that did not align with these databases were taken for further analysis. We used reference-based transcriptome assembly algorithms Hisat2 v2.1.0 (Kim et al., 2015); Cufflinks v2.2.1 (Trapnell et al., 2010) and Cuffdiff v2.2.1 (Trapnell et al., 2013) to identify differentially expressed transcripts. We used Hisat2 (-q -p 8 -min-intronlen 50 -max-intronlen 250000 -dta-cufflinks -new-summary -summary-file) to align the reads back to the dd_Smes_G4 (Grohme et al., 2018) assembly of S. mediterranea genome. Around 87–90% of reads were mapped to the dd_Smes_G4 genome. We used Samtools (http://www.htslib.org/) to obtain sorted bam files. The mapped reads were assembled using Cufflinks (-p -o -b -u -N -total-hits-norm -G) with the most recent and well-annotated SMEST transcriptome as reference. We used cuffmerge to merge the gene list across different conditions. We identified differentially expressed genes using Cuffdiff module (-p -o./-b -u -N -total-hits-norm -L) and considered genes with adjusted P<0.05 as the significance cut-off. Genes with a significant P-value and at least 2-fold up- or down-regulation were considered for further analysis. We did pathway analysis and gene ontology analysis for these selected up- and down-regulated transcripts using GSEA (Mootha et al., 2003; Subramanian et al., 2005). We used a customized Perl script for all the analyses used in this study (available upon request). We used R ggplot2 (Wickham, 2009), pheatmap and CummeRbund (Trapnell et al., 2012) library for plotting. Different planarian cell-type markers were obtained from the available single-cell transcriptome data (Fincher et al., 2018).

We used a single-cell transcriptome dataset published from Peter Reddien's laboratory (Fincher et al., 2018; GEO accession number GSE111764) to extract the cells that express ets-1 (dd_v6_2092). We used the data matrix submitted in the sequence read archive (SRA) to extract only the cells that express ets-1. We found 3347 cells (out of 50,563 cells) expressing ets-1 (dd_v6_2092) transcripts in the single-cell transcriptome. We reanalyzed the single-cell data as described by Ross and colleagues (Ross et al., 2018; BioProject accession number PRJNA432445). We used Seurat (https://satijalab.org/seurat/), an R package, to analyze the single-cell transcriptome for the cells that express ets1 mRNA (Butler et al., 2018; Stuart et al., 2019). Based on the markers from single-cell transcriptome dataset (Fincher et al., 2018), we classified the uniform manifold approximation and projection (UMAP) clusters as cell types. Neoblast and cathepsin⁺ clusters were two major clusters with 1331 and 1511 cells. We used LogNormalize method of Seurat to normalize the dataset, which was further scaled (linear transformation) using Seurat. This scaled value was further log-transformed and plotted as a heatmap for genes of interest. We used R ggplot2 GMD and heatmap.2 to derive all the plots.

Graph Pad prism (version-8.0.1) was used for statistical analysis and plotting graphs. Inkscape (version-1.1) and Bio-render (App.biorender.com) was used for making schematics and figure panels. iji (Version-ImageJ2.1. o/1.53c) was used for all the analysis and image processing.

We would like to thank Dr Rajesh Ladher, Dr Tina Mukherjee, Dr RamKumar Sambasivam, Dr Srikala Raghavan, and Dr Nishan Shettigar for their discussion and valuable feedback on our manuscript. Thanks also to Ms Swathi Pavithran and Ms Divyeksha Baraiya for their help in proofreading the manuscript. We acknowledge the central imaging and flow cytometry facility (CIFF) and NGS facility (special thanks to Dr Awadhesh Pandit) at the BLiSc Bio-Cluster for their constant technical support. Special thanks to Prof. Kiyokazu Agata for the kind gift of anti-arrestin (VC-1) antibody. Monoclonal antibody for planarian muscle (6G10) was developed by Kelly G. Ross at Department of Biology, San Diego State University and was obtained from the Developmental Studies Hybridoma Bank maintained at Department of Biology, University of Iowa.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.259900.reviewer-comments.pdf.