DVL1 and DVL3 require nuclear localisation to regulate proliferation in human myoblasts

WNT signalling is essential for regulating a diverse range of cellular processes. In skeletal muscle, the WNT pathway plays crucial roles in maintenance of the stem cell pool and myogenic differentiation. Focus is usually directed at examining the function of central components of the WNT pathway, including β-CATENIN and the GSK3β complex and TCF/LEF transcription factors, in tissue homeostasis and cancer. Other core components of the WNT pathway though, are three dishevelled (DVL) proteins: membrane associated proteins that propagate WNT signalling from membrane to nucleus. Here we examined DVL function in human myogenesis and the muscle-related cancer alveolar rhabdomyosarcoma. We demonstrate that DVL1 and DVL3 are necessary for efficient proliferation in human myoblasts and are important for timely myogenic differentiation. DVL1 and DVL3 also contribute to regulation of proliferation in rhabdomyosarcoma. DVL1 or DVL3 must be present in the nucleus to regulate proliferation, but they operate through different protein domains: DVL3 requires the DIX and PDZ domains, while DVL1 does not. Importantly, DVL1 and DVL3 activity is independent of markedly increased translocation of β-CATENIN to the nucleus, normally a hallmark of active canonical WNT signalling.


Results
Expression profile of DVL isoforms during human skeletal myogenesis. We used immunolabelling to examine the expression dynamics of DVL1-3 in proliferating human immortalised C25 myoblasts (Fig. 1A), early differentiating myocytes (Fig. 1B) and differentiated multinucleated myotubes (Fig. 1C). While immunolabelling was weak, we could discern that localisation of the DVL isoforms differed. During proliferation the location of DVL1 was both cytoplasmic and nuclear, while DVL2 was detected clearly in the nucleus and DVL3 was predominantly cytoplasmic (Fig. 1A). This localisation of DVL1-3 was retained through myogenic differentiation, with the notable exception that DVL2 appeared mainly cytoplasmic during the later stages of differentiation ( Fig. 1B and C).
We also examined the expression profile of the DVL isoforms through myogenesis using RT-qPCR (Fig. 1D). mRNA levels of DVL1-3 remained constant in proliferating C25 myoblasts and through myogenic differentiation, although there appeared to be a (non-significant) trend towards increased DVL1 and DVL3 expression as differentiation progressed. Successful myogenic differentiation was confirmed by a significant decrease in expression of a proliferation marker (CD1), but significant increases of early (MYOGENIN) and late (MYHC) differentiation markers (Fig. 1D). three DVL isoforms are expressed in proliferating human myoblasts, we performed siRNA-mediated knockdown of each DVL individually. RT-qPCR was performed to ensure that knockdown was specific to the target DVL isoform. As expected, a significant decrease in levels of the targeted DVL isoform occurred in each case for healthy control C25, 16U and 54-6 human myoblasts ( Fig. 2A and Supplementary Fig. 1A). With exception of the targeted DVL isoform, no consistent changes in expression of the other DVL isoforms occurred in the three independent immortalised human myoblast lines. In the C25 cell line only, there was a potential increase in DVL1 expression when DVL2 was knocked down ( Fig. 2A and Supplementary Fig. 1A) so we did not consider this a consistent effect. Morphologically, C25 myoblasts appeared more rounded and smaller when DVL1 was knocked down, while DVL2 or DVL3 knockdown caused a more elongated phenotype compared with control (Fig. 2B). Similar phenotypes were seen in 16U ( Supplementary Fig. 1B) and 54-6 ( Supplementary Fig. 1C) myoblasts after DVL knockdown. Quantification of cell circularity and size in proliferating C25 cells revealed that knockdown of DVL1 resulted in a rounder myoblast morphology with a smaller area, while knockdown of either DVL2 or DVL3 caused a more elongated morphology, which was stronger with DVL3 knockdown (Fig. 2C).
To investigate if knockdown of DVL isoforms affected proliferation, we performed a 2 h EdU pulse. Knockdown of DVL1 or DVL3 led to significantly less incorporation of EdU, so a reduced proliferation rate, while DVL2 knockdown had no effect. This was a consistent observation across the three human myoblast lines C25, 16U and 54-6 ( Fig. 2B, D and Supplementary Fig. 1B, C).  MYH8 and MYH2) differentiation markers in C25 myoblasts on day (d) 1 in proliferation (d1P) and on days (d) 1, 2 and 3 of differentiation (d1D, d2D and d3D). Data is represented as mean ± SD. N = 3, statistically significant differences were assessed using a One-WAY ANOVA with Dunett's post hoc test, comparing each time-point to proliferation where an asterisk denotes p < 0.05 and three asterisks denote p < 0.001.

Figure 2.
Knockdown of DVL1 or DVL3 reduces proliferation in human myoblasts. (A) Gene expression levels of DVL1 (left), DVL2 (middle) and DVL3 (right) in proliferating C25 myoblasts after knockdown of each isoform for 48 h. N = 3 biological replicates. Data is represented as mean ± SD with significant differences calculated using a One-Way ANOVA with Dunett's post-hoc test, comparing each group with the control (SiCtrl), where an asterisk denotes p < 0.05, two asterisks p < 0.01 and 3 asterisks p < 0.001. (B) Representative images of proliferating C25 myoblasts after knockdown of each DVL isoform for 48 h, immunolabelled for β-TUBULIN (green) to show cell morphology, with nuclei counterstained with DAPI (blue). Incorporated EdU is also visualised (red). Scale bar represents 100 µm. (C) Quantification cell circularity (0 equates to a straight line, 1 to a perfect circle) and size in proliferating C25 myoblasts after knockdown of each DVL isoform for 48 h, displayed as violin plots. Red line demarcates mean. Statistically significant differences were assessed using a One Way ANOVA with a Dunett's post hoc test, comparing each knockdown individually to the control (SiCtrl). N = 135 cells from 3 independent experiments. Three asterisks denote p < 0.001. (D) Quantification of the percentage of human C25, 16U and 54-6 myoblasts that had incorporated EdU after knockdown of each DVL isoform. More than 200 nuclei were analysed for each of N = 3-6 wells. (E) Crystal violet staining of C25 myoblasts after DVL knockdown after 24 or 96 h of proliferation. (F) Quantification of crystal violet incorporation in C25 myoblasts after DVL knockdown after 24, 72 and 96 h of proliferation. N = 4. Data is represented as mean ± SD. Statistical differences were assessed using a One-Way ANOVA with Dunett's post hoc test, comparing each knockdown group with the control, for each time-point, where two asterisks denote p < 0.01 and three asterisks p < 0.001. www.nature.com/scientificreports/ A crystal violet stain of C25 myoblasts after knockdown of each DVL isoform also demonstrated less cell density when DVL1 or DVL3 levels were reduced, compared to DVL2 or controls (Fig. 2E). Absorbance of crystal violet as a proxy for cell viability was also measured in C25 myoblasts after knockdown of each DVL isoform at multiple time-points, and revealed a reduced amount of cells after knockdown of DVL1 or DVL3 for 96 h, compared to the control (Fig. 2F). DVL1 and DVL3 are required for myogenic differentiation. When myoblasts enter myogenic differentiation, they exit the cell cycle and so we investigated whether the decrease in the proliferation rate caused by knockdown of DVL1 or DVL3 was a consequence of precocious differentiation. MYOGENIN drives early myogenic differentiation and is expressed at high levels in myocytes as they prepare to fuse to form multinucleated myotubes. Knockdown of each DVL isoform was followed by immunolabelling for MYOGENIN. This revealed that early differentiating cultures (day 1 of differentiation in C25 and day 2 of differentiation in 16U) exhibited a marked reduction in the percentage of myocytes containing MYOGENIN after knockdown of DVL1 or DVL3, while knockdown of DVL2 had no effect ( Fig. 3A and B, Supplementary Fig. 2). Thus knockdown of DVL1 or DVL3 not only significantly reduces the proliferation rate, but also interferes with entry into the myogenic differentiation program.
Terminal differentiation of C25 myoblasts was also reduced after knockdown of DVL1 or DVL3, as determined through quantifying the fusion index (percentage of nuclei in MyHC-expressing myotubes with 2 or more nuclei). The phenotype was most pronounced after DVL1 knockdown, when barely any multinucleated myotubes formed, but was also marked after knockdown of DVL3 ( Fig. 3C and D). To account for the reduced proliferation caused by DVL1 or DVL3 knockdown, cells were seeded at equal density before differentiation and quantification of nuclei per field confirmed that there was no significant difference in numbers (Fig. 3E). Thus, DVL1 and DVL3 are required for myogenic differentiation.
Constitutive overexpression of DVL1 or DVL3 increases myoblast proliferation. As knockdown of DVL1 or DVL3 had such a drastic effect on proliferation, we wanted to understand if constitutive expression would have an opposite effect. We cloned each DVL isoform into a pUltra lentiviral backbone also encoding eGFP separated by a self-cleaving P2A site, so that the presence of GFP is correlated with that of the DVL isoform. We used these lentiviruses and FAC sorting to generate stably overexpressing human C25 myoblast lines termed DVL1 + , DVL2 + or DVL3 + . RT-qPCR confirmed that each DVL isoform was specifically overexpressed, and that overexpression of each DVL isoform did not affect expression of the other two (Fig. 4A).
We did not observe any overt morphological changes after constitutive overexpression of any DVL isoform in C25 (Fig. 4B) or 16U ( Supplementary Fig. 3), in contrast to the effects of DVL1 or DVL3 knockdown ( Fig. 2B and Supplementary Fig. 1B, C). We assessed the proliferation rate using a 2 h EdU pulse, as per the knockdown experiments. Constitutive DVL1 or DVL3 overexpression in DVL1 + or DVL3 + C25 or 16U myoblasts significantly increased the proliferation rate in both myoblast lines, while DVL2 overexpression in DVL2 + myoblasts had no effect (Fig. 4B, C and Supplementary Fig. 3). DVL 1 or DVL3-mediated increase in proliferation is independent of WNT ligand stimulation. As DVL proteins are a core component of WNT signalling, we next tested if DVL-modulated effects on proliferation were a response to increased WNT signalling. LiCl inactivates GSK3β, allowing non-phosphorylated β-CATENIN to translocate to the nucleus and activate canonical WNT signalling [41][42][43] . We performed an EdU assay on proliferating C25 myoblasts that had been stimulated for 24 h with 5 mM LiCl prior to the 2 h EdU pulse. Activation of canonical WNT signalling significantly increased the proliferation rate of human C25 myoblasts, with the percentage of myoblasts incorporating EdU rising from 29 to 36.7% (Fig. 5A), as previously observed in murine satellite cells 44 and immortalised C2C12 myoblasts 45 .
Since LiCl activates canonical WNT signalling downstream of DVLs, we next assessed if the same pro-proliferative effect occurred through stimulation with WNT ligands WNT7a or WNT3a. We first isolated protein from C25 myoblasts stimulated with 10 ng/ml WNT7a or WNT3a for 24 h and western blotted for total β-CATENIN and active non-phosphorylated (Ser33/37Thr41) β-CATENIN in the nuclear faction. Total β-CATENIN was unchanged between the WNT-stimulated and untreated controls. Phosphorylation of β-CATENIN catalyses degradation of the protein, and so only stabilised dephosphorylated β-CATENIN translocates into the nucleus. Stimulation with WNT3a increased levels of active non-phosphorylated β-CATENIN in the nuclear fraction ( Fig. 5B, with the N = 3 full membranes shown in Supplementary Fig. 4). Given that LiCl stimulation increased proliferation, and WNT3a stimulation increased active nuclear β-CATENIN, it was unexpected that stimulation with WNT3a (or WNT7a) did not increase proliferation (Fig. 5C). Nuclear fractioning of C25 DVL1 + or DVL3 + myoblasts stably overexpressing DVL1 or DVL3 did not reveal an increase in active β-CATENIN, despite enhanced proliferation (Fig. 5D, with the N = 3 full membranes shown in Supplementary Fig. 5). Thus, increased proliferation via DVL1 or DVL3 appears governed through mechanisms other than facilitating translocation of β-CATENIN to the nucleus. DVL1 and DVL3 require the NLS signal to enhance proliferation in human myoblasts. Even though levels of active non-phosphorylated β-CATENIN were not changed by overexpression of DVL1 or DVL3, increased DVL expression may still enhance canonical WNT signalling in a less traditional way. DVL proteins in the nucleus form a complex with c-JUN and β-CATENIN on the promoter of WNT target genes, indicating that they can function as co-transcription factors 46 . Each DVL isoform contains a conserved nuclear localisation signal (Ile * Leu Thr) after the PDZ domain (Fig. 6A). We mutated this nuclear localisation signal to Ala * Gly Ala in DVL1 and DVL3 (since manipulation of DVL2 expression had no effects) (Fig. 6A)   , DVL2 (middle) and DVL3 (right) in proliferating human C25 DVL1 + , DVL2 + and DVL3 + myoblasts with stable lentiviral-mediated overexpression of each isoform. N = 4 biological replicates. Significant differences were calculated using a One-Way ANOVA with Dunett's post-hoc test, comparing each group with the control (pUltra) where an asterisk denotes p < 0.05, two asterisks denote p < 0.01 and three asterisks is  (Fig. 6B). Additionally, while C25 DVL1-mNLS + myoblasts had generally normal morphology, overexpression of DVL3-mNLS had drastic effects: attachment of DVL3-mNLS + myoblasts after seeding was compromised, and the cells that did attach were either small and round, or flattened with uncharacteristically long protrusions (Fig. 6B, C).

DVL3 operates via the DIX and/or PDZ domain to enhance proliferation in human myoblasts.
In addition to their function at the cell membrane/in the cytoplasm of propagating canonical and non-canonical WNT signalling, the DIX and PDZ domains of DVL proteins are also implicated in the nuclear functions of DVL. DVL1 and DVL3 mutants were generated lacking each of the three conserved domains (DIX, PDZ and DEP) (Fig. 7A), and the effect of their overexpression on myoblast proliferation was assessed after lentiviral transduction and selection of stable myoblast lines. Specific deletion of each domain individually in DVL1-ΔDIX + , DVL1-ΔPDZ + or DVL1-ΔDEP + C25 myoblasts had no effect on the function of the protein in enhancing proliferation, as per overexpression of wildtype DVL1 in DVL1 + myoblasts (Fig. 7B, C). However, the pro-proliferative effect of DVL3 overexpression in DVL3 + myoblasts relied on the DIX and/or PDZ domains, as removal of either in DVL3-ΔDIX + or DVL3-ΔPDZ + myoblasts caused the proliferation rate to return to control levels ( Fig. 7B, C). Overexpression of DVL3 lacking the DEP domain in DVL3-ΔDEP + myoblasts increased their proliferation, as per overexpression of wild type DVL3 (Fig. 7B, C).

Knockdown of DVL1 or DVL3 reduces proliferation in alveolar rhabdomyosarcoma cells. DVL
expression is commonly increased in many types of cancer 41,47-49 . Rhabdomyosarcoma, a childhood onset cancer  and the nuclear export sequence (NES), together with the changes to key amino acid residues to destroy the NLS (red text). (B) Representative images of proliferating C25 DVL1 + , DVL1-mNLS + , DVL3 + , DVL3-mNLS + and control (Ctrl) myoblasts, with incorporated EdU (red), immunolabelled for GFP (green) from the lentivirus (correlated with DVL expression) and DAPI nuclei counterstain (blue). EdU pulse was performed 24 h after seeding stably overexpressing C25 myoblast lines. Scale bar represents 100 µm. Quantification was performed for 4 biological replicates, and significant difference was calculated using a Oneway ANOVA with Dunnett's post hoc test, comparing each group with the control, where an asterisk denotes p < 0.05 and 3 asterisks denotes p < 0.001. (C) Representative images of DVL3 + (left) or DVL3-mNLS + (right) C25 myoblasts. Red arrows point to an unusually round, and an uncharacteristically elongated, DVL3-mNLS + myoblast.  in RH30 ARMS cells exposed to high-serum medium, co-immunolabelled for β-TUBULIN (red) with nuclei counterstained with DAPI (blue). (C) Gene expression profiles for DVL1-3, the proliferation marker CD1 and the early differentiation marker MYOGENIN in RH30 cells exposed to high or low-serum medium. N = 3 independent biological replicates. Significant differences were assessed using a students unpaired two tailed T-test, comparing high-serum to low-serum culture conditions where two asterisks denote p < 0.01. (D) Representative images of proliferating RH30 cells that have incorporated EdU (red) after individual knockdown of each DVL isoform for 48 h, with nuclei counterstained with DAPI (blue). Scale bar represents 100 µm. Quantification of EdU containing cells as a percentage of total nuclei after knockdown of each DVL isoform. Statistically significant differences were calculated using a One-Way ANOVA with Dunett's post-hoc test, comparing each group with the control (SiCtrl). A significant difference is denoted via two asterisks where p < 0.01. More than 200 nuclei were analysed for each of N = 3 biological replicates. www.nature.com/scientificreports/ DVL protein was readily detectable using immunofluorescence in proliferating RH30 cells maintained in high serum medium (Fig. 8B). DVL1 and DVL2 were localised to the nucleus, while DVL3 appeared predominantly cytoplasmic (Fig. 8B). Low serum culture conditions induce differentiation in muscle cells, and similarly, RH30 cells increased MYOGENIN expression but CD1 expression did not decrease significantly, showing that they remain proliferative (Fig. 8C). None of the three DVL isoforms showed any significant change in expression between high and low serum conditions (Fig. 8C).
After establishing that the DVL proteins were present in RH30 cells, we performed siRNA-mediated DVL knockdown and quantified the proliferation rate after a 2 h EdU pulse delivered 48 h later. Knockdown of DVL1 or DVL3 significantly reduced the proliferation rate of RH30 cells compared to siRNA control (SiCtrl), while knockdown of DVL2 did not affect the percentage of cells that had incorporated EdU (Fig. 8D). This was the same effect that DVL1 or DVL3 knockdown had on the proliferation rate of human myoblasts (Fig. 2). DVL1 and DVL3 require nuclear localisation to enhance proliferation in alveolar rhabdomyosarcoma cells. Even though RH30 cells are already highly proliferative, lentiviral-mediated stable overexpression of DVL1 or DVL3 further increased their proliferation rate, as shown by the higher percentage of cells that incorporated EdU after a 2 h pulse (Fig. 9A). To determine if the enhanced proliferation was dependent on the nuclear localisation of DVL1 or DVL3, we again used our DVL1-mNLS or DVL3-mNLS constructs, with the mutated NLS sequence. Constitutive overexpression in DVL1-mNLS + or DVL3-mNLS + RH30 cells prevented the increase in the proliferation rate observed when wild-type DVL1 or DVL3 were overexpressed (Fig. 9A). Interestingly, mutation of the NLS in DVL3-mNLS + RH30 cells actually reduced the proliferation rate to below that of the control (Fig. 9A). Overexpression of the DVL3-mNLS also led to a reduced number of DVL3-mNLS + RH30 cells compared to the other conditions (Fig. 9A). Since identical numbers of RH30 cells were seeded, this suggests that overexpression of DVL3-mNLS drastically reduced the proliferation rate, and/or prevented ARMS cells from attaching/caused cell death.
As the DVL3-mNLS mutant had a more pronounced effect compared with the DVL1-mNLS mutant, we stably overexpressed our DVL3-ΔDIX, DVL3-ΔPDZ and DVL3-ΔDEP mutants in RH30 cells. Deletion of the DIX or PDZ domain prevented DVL3 from increasing the proliferation rate in DVL3-ΔDIX + or DVL3-ΔPDZ + RH30 cells (Fig. 9B). In contrast, deletion of the DEP domain in DVL3-ΔDEP + RH30 cells actually enhanced proliferation, even over that achieved by overexpression of the full length DVL3 protein (Fig. 9B).

Discussion
DVL proteins are predominantly known for their role as scaffolding proteins, acting between Frizzled receptors and the β-CATENIN destruction complex in WNT signalling. During active WNT signalling, DVL proteins are phosphorylated to inhibit the destruction complex, allowing non-phosphorylated β-CATENIN to translocate to the nucleus 11,12 . However, a dual role for DVL at both the cell membrane/in cytoplasm and nucleus has been established 52 . The presence of DVL in the nucleus is necessary for β-CATENIN-dependent downstream gene expression 15 , including through activation of TCF/LEF-mediated transcription together with c-JUN.
In this study, we show that DVL1 and DVL3, but not DVL2, regulate proliferation in human myoblasts and are necessary for successful myogenic differentiation. Indeed, DVLs are requisite for efficient myoblast function as when either DVL1 or DVL3 are knocked down, both the proliferation rate, entry into myogenic differentiation and fusion are compromised. We also report that DVLs are required for maximal proliferation in ARMS cells: consistent with observations that DVL overexpression is associated with higher proliferation, invasiveness and aggressiveness in breast cancer 47,53 , prostate cancer 54 , primary cervical squamous cancer 55 and high-grade gliomas 56 . Curiously though, it is the actions of DVL1 or DVL3 in the nucleus that facilitates these effects on proliferation in myoblasts and ARMS.
DVL proteins exhibit head to tail dimerisation via the DIX domain and form large protein puncta 1-2 µM in size when overexpressed 57,58 . That puncta assemble, strongly suggests that DVL proteins can dimerise. Additionally, upon WNT stimulation, DVL2 concentration at the plasma membrane increases and the protein oligomerises 28 . We found that mutating the NLS sequence reduced the cell proliferation rate, showing that nuclear localisation is important. This may be a direct effect of the mutant NLS on DVL1 or DVL3, and/or the mutant DVLs may dimerise/oligomerise with endogenous wildtype DVL, preventing it from operating. This would explain why there is a reduction in the proliferation rate, below even the control level.
Interestingly, we did not observe an increase in nuclear β-CATENIN in response to DVL overexpression, suggesting that the pro-proliferative effects of DVL1 or DVL3 are not only due to nuclear translocation of β-CATENIN. However, nuclear β-CATENIN is present in unstimulated cells, and so the pool of nuclear β-CATENIN may already be sufficient to drive higher levels of WNT signalling when levels of co-activators such as DVL increase. Since DVLs facilitate β-CATENIN dependent transcription via TCF/LEF transcription factors within the nucleus 15,16,52 , overexpressing DVLs could further increase the transcriptional output of canonical WNT signalling. ChIP-Seq to further understand the putative role of DVLs as transcription factor/co-factors would be informative.
A consistent finding was that manipulating either DVL1 or DVL3 had similar effects, while manipulation of DVL2 did not measurably affect myoblast behaviour. Similarly, knockdown of DVL1 or DVL3 in non-muscle cell lines HEK293T (human embryonic kidney cells) or P19 cells (embryonic carcinoma cell line) has a stronger effect on canonical WNT signalling than DVL2 knockdown 26 . The DIX and PDZ domains of DVLs are involved in propagation of canonical WNT signalling (reviewed in 33 ), which would explain why their deletion abrogated the effect of DVL3 overexpression. Deletion of individual domains does not inhibit nuclear localisation of DVLs however 15 , so while DVL3 lacking the DIX or PDZ domain may still localise to the nucleus, they are then incapable of functioning effectively in either myoblasts or ARMS cells. However, the DEP domain is necessary for Scientific Reports | (2022) 12:8388 | https://doi.org/10.1038/s41598-022-10536-x www.nature.com/scientificreports/ membrane recruitment of DVL 59 , which likely explains why deletion of the DEP domain did not alter DVL3 function. Therefore, DVL3 regulates myoblast proliferation via its nuclear localisation and likely ability to then propagate canonical WNT signalling via its DIX or PDZ domains. In contrast to DVL3, provided that DVL1 is able to enter the nucleus, mutations of either the DIX, PDZ or DEP domains have no effects on its ability to enhance the proliferation rate in myoblasts. Thus, while manipulation of DVL3 or DVL1 led to the same biological outcome in myoblasts, they employ different mechanism. Two acetylated lysine residues in the PDZ and DIX domain of DVL1 also promote nuclear localisation (in addition to the NLS) and promoter binding 60 . If those residues can compensate for each other, then deletion of both amino acids/domains would be required to inhibit the pro-proliferative effect of DVL1 overexpression. In DVL3, lysine residues do not occupy these two positions, which may explain the different mechanism of action. Representative images of stable proliferating DVL1 + , DVL1-mNLS + , DVL3 + , DVL3-mNLS + and control (pUltra) RH30 ARMS cells with incorporated EdU (red) and a DAPI nuclear counterstained (blue). Scale bar represents 100 µm. Quantification was performed for 3 biological replicates, and a significant difference was calculated using a student's T-test, where an asterisk denotes p < 0.05 comparing the two groups connected with a line and # denotes p < 0.05 compared to control. (B) Representative images of proliferating DVL3 + , DVL3-ΔDIX + , DVL3-ΔPDZ + , DVL3-ΔDEP + or control (pUltra) RH30 cells with EdU incorporation visualised (red) and a DAPI nuclear counterstained (blue). Scale bar represents 100 µm. Quantification was performed for 4 biological replicates, and significant difference calculated using a One-Way ANOVA, comparing each group against the control, where an asterisk denotes p < 0.05, and three asterisks denotes p < 0.001. www.nature.com/scientificreports/ Manipulation of DVL2 had no effects on proliferation in myogenic or ARMS cells. DVL2 is the most abundantly expressed DVL isoforms in multiple cell lines, such as murine F9 and P19 cells and human HEK293 cells 26 . Given that siRNA-mediated knockdown had a maximum efficiency of 85%, it is possible that the remaining pool of DVL2 was still sufficient to maintain its function. In murine skeletal muscle 61 and immortalised muscle C2C12 cells 24,27 , protection of DVL2 from autophagy increases canonical WNT signalling which in turn increases differentiation. Those studies show significant effects during muscle regeneration and differentiation, but since our experiments were predominantly restricted to proliferation, it is possible that the effect of DVL2 manipulation would only become evident in mature muscle, and this late stage of myogenesis cannot be reliably modelled in our 2D cell culture system.

Scientific Reports
Co-immunoprecipitation to identify DVL interactors in the nucleus would further pinpoint mechanistic differences between DVL1 and DVL3. Additionally, considering the importance of the PDZ and DIX motifs for proliferation in ARMS cells, specific inhibitors could be tested as part of potential therapies for this difficult to treat sarcoma.

Conclusions
We demonstrate the role of DVL1 and DVL3 in proliferation of human myoblasts and ARMS cells. Knockdown of either DVL1 or DVL3 significantly reduced the proliferation rate, while constitutive expression of either DVL1 or DVL3 enhanced proliferation. DVL1 or DVL3 knockdown also inhibited myoblasts from entering myogenic differentiation and forming multinucleated myotubes. This process was independent of WNT ligand stimulation, as addition of either WNT3a or WNT7a did not change the cell proliferation rate. A functional nuclear localisation signal was essential to maintain the pro-proliferative effect of either DVL isoforms, however the mechanism of how DVL1 or DVL3 overexpression increases proliferation diverges between the two isoforms. Removal of the DIX, PDZ or DEP domain from DVL1 did not interfere with its capacity to increase proliferation, while removal of the DIX or PDZ domain from DVL3 abrogated its ability to enhance cell proliferation.

Materials and methods
Cell lines. Immortalised human myoblast lines C25 and 54-6 were obtained from the Institute de Myologie (France) 62,63 , while the 16U line was from the UMMS Wellstone Centre for FSHD (USA). The RH30 ARMS cell line (CVCL_0041) was from the Institute of Cancer Research, London 64 . All cell lines were regularly tested for mycoplasma, and each batch were cultured for less than 3 weeks.
For WNT ligand stimulation experiments, cells were plated in 6 well plates and 24 h later were stimulated with 10 ng/µl WNT3a or WNT7a (final concentration) and protein harvested 24 h later.
For overexpression, 20 000 cells were transduced with 0.5 ml supernatant containing viral particles. GFP expression was observed 24-48 h later and cell lines were passaged at least 2 times before being plated for experiments. Transduction was performed in myogenic proliferation medium except for RH30 cells, for which DMEM based proliferation medium was used.
Immunolabelling. Cells were fixed with 4% paraformaldehyde (PFA)/PBS for 10 min, permeabilized with 0.05% Triton/PBS for 10 min and non-specific antibody binding blocked with 10% goat serum/PBS for 30 min. Cells were then incubated with the primary antibody in PBS/1% goat serum (Table 2) at 4 °C overnight followed by three PBS washes and incubation with the secondary antibody (Invitrogen, 1:500) for 1 h at room temperature. After further PBS washes, nuclei were stained with DAPI (1:1000) and mounted for viewing and imaging on an aging Zeiss AxioVert 200 M epifluorescence microscope with a Zeiss AxioCam HRm and AxioVision 4.4 software (Zeiss, Jena, Germany).
For proliferation rate analysis, cells were pulsed with EdU for two hours prior to fixation. EdU incorporation was visualised using the Click-iT EdU Imaging Kit (Life Technologies) as per manufacturer's instructions. EdU positive nuclei were counted and expressed as the percentage of total nuclei within the field.  Protein extraction and western blot. Protein and subcellular fractioning were performed according to the ABCAM protocol. In brief: plated cells were washed with ice cold PBS, scraped, lysed in fractioning buffer (20 mM HEPES, 10 mM KCl, 2 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA and freshly added 1 mM DTT and PI Cocktail III 1:200) and incubated on ice for 15 min. Samples were passed through a 27G needle until cells were lysed and incubated on ice for 20 min. Samples were centrifuged at 720 g for 5 min and the supernatant containing the cytoplasm fraction collected. The nuclear pellet was washed in 500 µl fractioning buffer, passed through a 23G needle 10 times and centrifuged for 10 min at 720 g. The supernatant (containing the nuclear Table 1. Primer sequences for cloning DVLs into the pUltra backbone.  www.nature.com/scientificreports/ proteins) was collected. Supernatants containing nuclear or cytoplasmic proteins were mixed with 4 × Laemmli buffer, boiled for 5 min at 95 °C and frozen at − 20 °C for storage. Protein and 5 µl precision plus protein standards dual colour ladder (BioRad, #161-0374) were loaded on a 4-20% precast gel (BioRad, #4561094), run for 1 h at 60 V and transferred to nitrocellulose membrane at 70 V for 1 h. The membrane was cut to immunolabel for β-CATENIN and controls VINCULIN (Cell Signaling, #13901) or TBP (Cell Signaling, #8515). The strip containing β-CATENIN was first probed for active non-phospho β-CATENIN (Cell Signaling, #8814S), then stripped (ThermoFisher, 21059) and re-probed for total β-CATENIN (Cell Signaling, #L87A12). VINCULIN was used as housekeeper for the cytoplasmic fraction and TBP as housekeeper for the nuclear fraction. Band intensities were quantified and normalised to the housekeeper intensity.
RNA-seq data analysis. Hayes