Functional Dissection of Pax3 in Paraxial Mesoderm Development and Myogenesis§


  • Author contributions: A.M.: designed and performed research, analyzed data, and wrote the manuscript; E.S., F.R., and P.B.: performed research and analyzed data; R.C.R.P.: contributed with interpretation of the data and wrote the paper.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLSEXPRESS October 18, 2012.


The paired box transcription factor Pax3 is well-known as a major regulator of embryonic myogenesis. Before Pax3 expression becomes restricted to the dermomyotome, this transcription factor is also expressed in the developing somites. The role of Pax3 at this early stage is unclear, in particular because of the scarce frequency of Pax3-positive cells in the early mouse embryo. Inducible gene expression in embryonic stem cells (ESCs) represents an excellent tool to overcome this limitation, since it can provide large quantities of otherwise rare embryonic populations expressing a factor of interest. Here we used engineered mouse ESCs to perform a functional analysis of Pax3 with the aim to identify the molecular determinants involved in the early functions of this transcription factor. We find that Pax3 induction during embryoid body differentiation results in the upregulation of genes expressed in the presomitic and somitic mesoderm. Moreover, we show that paraxial mesoderm induced by transient expression of Pax3 is not irreversibly committed to myogenesis rather requires sustained Pax3 expression. Using a series of deletion mutants of Pax3, which differentially affect its transcriptional activity, we map protein domains necessary for induction of paraxial mesoderm and induction of the myogenic program. The paired, homeo-, and transcriptional activation domains were each required for both processes, however, the paired-c-terminal RED domain showed a paraxial mesoderm-specific activity that was dispensable for myogenesis. These findings demonstrate and provide mechanistic insight into an early role for Pax3 in the generation of paraxial mesoderm. STEM Cells2013;31:59–70


During development, mesoderm arises from cells that ingress through the primitive streak, a process that is regulated by gradients of morphogens that establish the body plan axes. Mesoderm is initially patterned into four general compartments: the axial mesoderm (notochord), the paraxial mesoderm (somites), the lateral plate mesoderm (cardiac and hemato-endothelial), and the extraembryonic mesoderm (reviewed by [1]). Commitment toward one of these compartments depends on both the time of ingression in the primitive streak and position of the cells, which defines the morphogens cells will be exposed to, ultimately resulting in the activation of specific transcription factors. After specification, paraxial mesoderm undergoes re-epithelialization as presomitic mesoderm (PSM) undergoes the process of segmentation into somites. Somitogenesis is regulated by morphogen gradients of retinoic acid and fibroblast growth factor 8 (FGF8) together with the synchronous oscillation of cells in the PSM that results in the cyclic expression of Notch pathway genes, that is, Lunatic Fringe (Lnfg) and Hairy/Enhancer of split (Hes). Notably, disruption of Notch signaling results in defective somitogenesis (reviewed by [2]). This process proceeds in an anterior to posterior direction and establishes alternating bands of cells that ultimately form an epithelial structure. After their formation, epithelial somites are patterned dorsally and ventrally to originate, respectively, the dermomyotome and sclerotome [3].

Pax3, a member of the paired domain family of transcription factors, is an important regulator of myogenesis (reviewed in [4]). During embryogenesis, Pax3 is expressed in the dorsal neural tube, in the posterior PSM, and in the developing somites, where expression becomes restricted to the epaxial and hypaxial regions of the dermomyotome. Ablation or mutation of Pax3, as in the case of the splotch mice [5], is characterized by the absence of limb muscles. In these embryos, muscle progenitors do not delaminate from the hypaxial dermomyotome, and additional developmental abnormalities arise in neural crest cell derivatives. Similarly, mutation of the human PAX3 gene is observed in patients affected by Waardenburg syndrome, which are characterized by defects in pigmentation, hearing, and limb musculature. Importantly, Pax3 regulates the expression of the hepatocyte growth factor (HGF) receptor c-Met, which is necessary for the delamination/migration of the muscle progenitors toward the limb bud [6]. Another well-known Pax3 target gene is the myogenic factor Myf5, which is induced directly by Pax3 in the hypaxial somite and limb [7, 8]. Direct targets of Pax3 also include the FGF receptor 4 (Fgfr4), the Forkhead family transcription factor FoxC2, and the Double-sex homolog Dmrt2 [9–11].

The transcriptional activity of Pax3 depends on the presence of two independent DNA binding domains, the paired domain and the homeodomain, which are conserved across evolution [12, 13]. Pax3, its homologous Pax7, and other Pax-family members also share a short sequence named the octapeptide, which has been proposed to bind the corepressor Groucho [14]. The carboxy-terminal transactivation domain is less characterized and less conserved. Several studies have indicated that Pax3 is not a strong transactivator, however, these used immortalized cell lines, which may not express relevant cofactors and clearly do not reproduce the early events that occur in the mouse embryo [8, 15–18]. However, expression of Pax3 in the P19 embryonal carcinoma cell line [19] or in mouse embryonic stem cells (ESCs) [20] is able to induce the myogenic program [19, 20], supporting the importance of an appropriate cellular model to investigate embryonic myogenesis. mESCs represent a powerful tool to study early developmental mechanisms particularly due to the possibility to manipulate their genome and to adapt their culture conditions [21–23]. Taking advantage of this system, we investigated the effect of Pax3 expression in differentiating mESCs, focusing on the characterization of domains and how they affect Pax3 activity on paraxial mesoderm and myogenesis. Through the generation and characterization of a mutant panel of ESC lines, we show that (a) induction of Pax3 in embryoid body (EB) cultures recapitulates the formation of the embryonic myotome, through a process involving upregulation of somite markers; (b) continual induction of Pax3 is required to induce the final commitment of these cells to the myogenic lineage, and (c) the paired-c-terminal domain of Pax3 plays an important role specific to paraxial mesoderm.


Plasmids and Cell Line Generation

The recombination plasmid p2lox Pax3 was obtained by subcloning the blunted EcoRI/XhoI DNA fragment encoding Pax3 (from pSPORT Pax3, Open Biosystems-Thermo Scientific, Lafayette CO) into p2lox vector (described in [24]) digested with EcoRV and SmaI. The p2lox Pax3 3xFlag was generated in a two-step procedure. First, Pax3 was polymerase chain reaction (PCR)-amplified to remove the stop codon with the following primers FW: ACA GAATTCATGACCACGCTGGCCG; RV: CAGGCGGCCGC TGCAATATCTGGCTTGAG and subcloned in the p2lox vector using EcoRI/NotI. Second, the sequence encoding the 3xFlag was inserted in the NotI site of the p2lox Pax3 (without stop codon) by ligating two annealed oligos (FLAG-FW and -RV) encoding the tag sequence. FLAG-FW: GGCCCTCGAGGACTACAAAGACCATGACGGTGATTATA AAGATCATGACATCGACTACAAGGATGACGATGACAAG TAGGC; FLAG-RV: GGCCGCCTACTTGTCATCGTCATC CTTGTAGTCGATGTCATGATCTTTATAATCACCGTCATG GTCTTTGTAGTCCTCGAG. Deletion mutants for both Pax3 and Pax3-3xFlag were generated using the Quickchange Mutagenesis kit (Stratagene, La Jolla, CA, following the manufacturer's instructions. The list of oligos used is reported in Supporting Information Table S1. The p2lox-Ires-GFP plasmid was generated by cloning of the blunted EcoRI-BsrGI DNA fragment encoding Ires-GFP (from pSAM2-ires-GFP—described in [25]) into p2lox vector digested with EcoRV and SmaI. The p2lox Pax3 Ires-GFP was generated by insertion of the blunted EcoRI-XhoI fragment (described above) into the p2lox-Ires-GFP vector digested with XhoI and blunted. Stable cell lines were generated by inducible cassette exchange by transfecting 20 μg of p2lox vector into the A2lox-CRE cell line [21] using the Mouse ES cell Nucleofector kit (Lonza, Allendale, NJ, Cells were selected with 300 μg/ml of Geneticin (Gibco, Grand Island, NY, as previously described [26]. Prior to the studies described here, all these mutants were sequenced to avoid PCR-mediated mutation (data not shown).

Cell Cultures

mESCs were cultured on irradiated mouse embryonic fibroblasts (MEFs) in knockout Dulbecco's modified Eagle's medium (DMEM) (Gibco) supplemented with 15% ES qualified FBS (Gibco), penicillin/streptomycin (Gibco), 2 mM Glutamax (Gibco), 0.1 mM β-mercaptoethanol, and 1,000 U/ml leukemia inhibitory factor (LIF) (Millipore, Billerica, MA,, at 37°C in 5% CO2. For differentiation, ESCs were trypsinized, resuspended in EB differentiation media (EBM), and preplated for 30 minutes on gelatin-coated dishes to remove MEFs. After counting, cells were diluted in EBM at 10,000 cells per milliliter and plated as drops of 10 μl in nonadherent 15-cm Petri dishes to induce EB formation. EBs were harvested after 48 hours and plated on nonadherent 10-cm dishes on a swirling rotator in EBM at 37°C in 5% CO2. EBM consisted of Iscove-modified DMEM (Gibco) supplemented with 15% ES qualified FBS (Gibco), penicillin/streptomycin, 2 mM Glutamax, 200 μg/ml iron-saturated transferrin (Sigma, St. Louis, MO), 4.5 mM monothioglycerol (Sigma), and 50 μg/ml ascorbic acid (Sigma). Doxycycline (Sigma, D9891) was dissolved in sterile PBS at a concentration of 1 mg/ml and stored at 4°C. Induction of the transgene was achieved by adding the doxycycline stock to the culture media at a final concentration of 0.8 μg/ml. Sorted or unsorted cells obtained from day 5 EBs were replated on gelatinized flasks cultured at 37°C in 5% CO2 in EBM for 5 days, followed by 1 day in DMEM (Gibco) supplemented with 2% horse serum (Hyclone, Logan, UT,, penicillin/streptomycin, and 2 mM glutamax to assess their differentiation into myotubes.

FACS Analysis

Day 5 EBs were harvested, washed twice with PBS, and then trypsinized for 1.5 minutes in a 37°C water bath with continuous shaking. Trypsin was inactivated by adding four volumes of phosphate buffered saline (PBS) supplemented with 10% FBS (PBSF), and cells were resuspended and filtered through a 70 μm strain to remove cell clumps. Cells were washed once with PBS and then incubated 5 minutes in PBSF in the presence of Fc Block (1 μl per 4 million cells, E-bioscience). Staining was performed by adding 0.5 μl of each antibody per one million of cells and incubating for 25 minutes on ice. We used a phycoerythrin (PE)-conjugated anti-mouse CD140a (PDGFRα) and an allophycocyanin (APC)-conjugated anti-mouse CD309 (FLK-1) (both from E-bioscience-San Diego, CA, Cells were washed twice with PBS and then resuspended in PBSF containing propidium iodide to exclude dead cells. Samples were analyzed and sorted using a FACSAria II (BD Biosciences, San Diego, CA,

RNA Isolation and Gene Expression

Cells or total EBs were resuspended in Trizol (Invitrogen, Carlsbad, CA, and processed following the manufacturer's instructions. RNAs from total EBs were retro-transcribed using Thermoscript (Invitrogen). RNAs from sorted cells were retro-transcribed using Superscript Vilo (Invitrogen). Gene expression analyses were performed using an amount of cDNA solution corresponding to 12.5 ng of starting RNA for each reaction. Quantitative reverse transcription PCR (qRT-PCR) was performed using TaqMan Universal PCR Master Mix and TaqMan probes (Applied Biosystems, Foster City, CA,

Western Blot and Immunofluorescence Analysis

Proteins were extracted from total EBs or single cells using RIPA buffer (150 mM NaCl, 50 mM Tris HCl pH 7.5, 1 mM EDTA, 1% Triton, 1% sodium deoxicholate, 0.1% SDS) supplemented of Protease inhibitors (complete, Roche Diagnostics, Basel, Switzerland, and quantified with Bradford reagent (Sigma). Protein samples were prepared in Laemmli buffer and loaded on gels for SDS-PAGE. Proteins were transferred on polyvinyl difluoride (PVDF) membranes (Millipore) for the detection with the indicated antibodies. Immunofluorescence staining was performed by fixing cells with 4% paraformaldehyde/PBS for 10 minutes at 4°C, then permeabilized with 0.1% Triton/PBS and blocked with 5% BSA/PBS before incubating with the primary antibodies. Samples were rinsed with PBS, blocked with 5% bovine serum albumine (BSA)/PBS, and then incubated with the secondary antibody. After washing, samples were mounted on the slides using Prolong Gold with DAPI (4′,6-diamidino-2-phenylindole - Molecular Probes, Eugene, OR, http://probes. Pictures were acquired with Axioimager M1 fluorescence microscope or LSM 510 Meta confocal microscope (Zeiss, Germany).

Antibodies used in this study included anti-MYOD (BD Biosciences, 554130), anti-MYOG (BD Biosciences, 556358), anti-PAX3 (C-term directed, Developmental Studies Hybridoma Bank, Iowa City, IA,∼dshbwww), anti-PAX3 (N-term directed, R&D System, Minneapolis, MN,, anti-GAPDH (Abcam, ab8245, Cambridge, U.K.,, anti-FLAG (Clone M2, Sigma), anti-cardiac Troponin I (anti-cTNI) (Abcam, ab47003), horseradish peroxidase (HRP)-conjugated anti-mouse and anti-rabbit IgG (Amersham Biosciences, Piscataway, NJ, http://www., Alexa488-conjugated anti-rabbit IgG, and Alexa555-conjugated anti-mouse IgG (Molecular Probes).

Statistical Analysis

Student's two-tailed t test was used to analyze differences between noninduced and dox-induced samples whereas differences between multiple samples were assessed by ANOVA.


Function of Pax3 in Paraxial Mesoderm and Myogenic Specification

Analyses of the molecular mechanisms involving Pax3 during embryogenesis have been limited by the availability of a cellular model able to mimic these early developmental stages. Because in vitro differentiation of mESC mimics the formation of most mesoderm derivatives, their scalable growth properties open a window into mechanistic studies to understand the activity of transcription factors during the development of rare embryonic precursors. An important aspect to be considered is that skeletal myogenesis is not properly recapitulated during the in vitro differentiation of ESCs into embryoid bodies (EBs), as evidenced by the very low levels of Pax3, Pax7, and muscle regulatory factors - MRFs - (Darabi et al. [23]) (Supporting Information Fig. S1A). We have previously shown that inefficient myogenesis can be bypassed by induction of Pax3 from day 2 of EB differentiation, when mesoderm first arises, resulting in the efficient activation of the myogenic program [20]. Interestingly, Pax3 induction also led to an increase in the frequency of PDGFRα+FLK-1 cells, a surface profile consistent with paraxial mesoderm [23]. To specifically elucidate whether this population induced by Pax3 is actually paraxial mesoderm, we made use of this Pax3 doxycycline-inducible ESC line and analyzed by qRT-PCR the expression levels of paraxial mesoderm and myogenic markers in total day 5 EBs, which were induced with doxycycline from day 2. As shown in Figure 1A and Supporting Information Figure S1B, Pax3 induction indeed results in an overall increase of all paraxial mesoderm markers tested, including Mesogenin1 (Msgn1), Delta-like1 (Dll1), FoxC1, Paraxis (Tcf15), Meox1, Tbx18, and Uncx4.1. Msgn1 and Dll1 are mainly expressed in the posterior PSM [27, 28], where they regulate somitogenesis [29]. Similarly, other mesodermal markers are upregulated after Pax3 induction. FoxC1 is necessary for the segmentation of the PSM and its differentiation [30], Paraxis (Tcf15) and Meox1 are both involved in the somite epithelialization [31, 32], and Tbx18 and Uncx4.1 represent, respectively, anterior and posterior somite markers [33, 34]. As expected, Pax3 induction resulted in upregulation of c-Met and Myf5 (well-known target genes of Pax3) as well as M-cadherin (Cdh15), MHox (Prrx1), and Nr2f2 (COUP-TFII) (Fig. 1A; Supporting Information Fig. S1B). Myf5 and M-cad are both expressed in the myotome [7, 35]. MHox is detected in the dorsal dermomyotome and in the limb buds [36], whereas Nr2f2 is required for limb development [37]. At this time point, expression of MyoD and Myf6 is not yet detectable (data not shown).

Figure 1.

Pax3 is necessary for the commitment of embryonic stem cells toward the paraxial mesoderm cell fate and skeletal myogenic lineage. (A): Paraxial mesoderm and myogenic markers are upregulated by Pax3 induction, as shown by quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis of total day 5 embryoid bodies. Results were normalized to Gapdh endogenous levels. Graphs represent the mean ± SD of at least three independent experiments. *, p < .05; **, p < .01; ***, p < .001. (B): Fluorescence-activated cell sorting (FACS) plots show green fluorescent protein (GFP) expression following a short window of dox induction (6 hours) in induced and noninduced iPax3-Ires-GFP and Ires-GFP (empty vector control) day 5 EBs. (C): Expression of Myf5, Meox1, and Dll1 is upregulated within 6 hours of Pax3 induction, as shown by quantitative RT-PCR analysis performed on sorted cells (B). *, p < .05; **, p < .01; ***, p < .001. Abbreviations: Dll1, Delta-like1; dox, doxycycline; FSC-H, Forward Scatter; GFP, green fluorescent protein; Msgn1, Mesogenin1.

The upregulation of genes that are important for paraxial mesoderm development in day 5 total EBs suggests that Pax3 also plays a role in the stages that precede myotome formation. To identify which of these genes might be directly regulated by Pax3, we generated a bicistronic inducible cell line expressing Pax3 and green fluorescent protein (GFP) (iPax3-Ires-GFP). After dox treatment, this cell line allows for the purification of cells that express Pax3 within 6 hours of induction. As shown in Figure 1B, GFP positive cells can be identified and sorted by fluorescence-activated cell sorting (FACS) and used for gene expression analysis. When compared to respective controls (Pax3-ires-GFP no dox and ires-GFP no/+dox), we were able to identify three genes upregulated in this short time window of Pax3 induction. As expected, Myf5, a well-characterized Pax3 target [7], was one of them (Fig. 1C). Among the paraxial mesoderm genes, we identified Meox1 and Dll1 as candidate targets of Pax3 (Fig. 1C). Given the importance of these genes in somite formation, it is possible that Pax3 positively regulates their expression. The role of Pax3 in the regulation of these genes is supported by previous observations: (a) a Pax3 binding site has been identified in rhabdomyosarcoma cells expressing PAX3-FKHR [16], and (b) expression of Dll1 is impaired in the somites of splotch homozygous mouse embryos at E9.5 [38].

These initial results prompted us to further investigate how markers of paraxial mesoderm and myogenesis were induced by Pax3 within different subpopulations of EBs. Taking advantage of the subdivision into lateral plate and paraxial mesoderm provided by the differential expression of PDGFRα and FLK-1 in the early stages of EB differentiation [23], we sorted these four populations (named P+F+, P+, F+, and NEG) from both Pax3-induced and noninduced day 5 EBs, as shown in Figure 2A. Since the purity of sorted cell populations is critical for the subsequent analysis, we reanalyzed each cell sorted fraction by FACS, which confirmed their purity above 95% (Supporting Information Fig. S1C). Gene expression analysis on the sorted cell populations revealed that the majority of the genes (Meox1, Myf5, c-Met, Prrx1, and Nr2f2) upregulated in the total EBs had a similar pattern of expression in the PDGFRα+FLK-1 (P+) fraction (Fig. 2B). Interestingly, this was not the case for genes known to play a role in regulating somitogenesis, such as Dll1, Msgn1, and FoxC1. These genes were instead upregulated in the PDGFRα+FLK-1+ (P+F+) and PDGFRαFLK-1+ (F+) fractions, suggesting that Pax3 may induce other mesodermal cells to gradually acquire a paraxial fate. As mentioned above, genes involved in the subsequent steps of differentiation (Meox1, Paraxis, Myf5, c-Met, Prrx1, and Nr2f2) were mainly upregulated in the PDGFRα+FLK-1 (P+) as well as the PDGFRα+FLK-1+ (P+F+) cell fractions. In the case of Paraxis, upregulation was also observed in the PDGFRαFLK-1+ (F+) fraction. As expected the expression of most of the paraxial/myogenic markers was significantly lower in the PDGFRαFLK-1 (NEG) fraction. All together, these data suggest that Pax3 can induce the paraxial/myogenic program mainly within cells already restricted to the mesodermal lineage. The PDGFRα+FLK-1 paraxial mesoderm cell population that emerges from EBs recapitulates somitogenic gene expression and is followed by differentiation toward the dermomyotome/myotome.

Figure 2.

Gene expression analysis on mesodermal sub-populations (A): Fluorescence-activated cell sorting (FACS) profile of total and subfractioned day 5 embryoid bodies (EBs) from dox-induced and noninduced cultures. Cells were stained with anti-PDGFRα and anti-FLK-1 antibodies, sorted, and reanalyzed to evaluate their purity. PDGFRα+FLK-1+ (P+F+), PDGFRα+FLK-1 (P+), PDGFRαFLK-1+ (F+), PDGFRαFLK-1 (NEG). The average frequency of cells in each subfraction is reported as percentage in the respective graph. (B): Quantitative reverse transcription polymerase chain reaction (RT-PCR) analysis of sorted day 5 EBs subfractions. Following Pax3 induction, paraxial mesoderm and myogenic genes were mainly upregulated in P+F+ and P+ fractions. Graphs represent the mean ± SD of at least three independent experiments. *, p < .05; **, p < .01; ***p < .001. Abbreviations: Dll1, Delta-like1; dox, doxycycline; Msgn1, Mesogenin1.

Maintained Expression of Pax3 Is Necessary to Induce the Myogenic Program

Based on the above results, we then aimed to determine whether the PDGFRα+FLK-1 cell fraction is fully committed to the myogenic lineage or represents a transitional plastic intermediate population. Our previous findings show that persistent induction of Pax3 drives the generation of a proliferating myogenic population that abundantly expresses Pax3 and Myf5, with upregulation of MyoD under differentiation conditions [20]. Therefore, we performed a time course analysis by inducing Pax3 at two time windows, either during EB differentiation (d2–d5) or during the monolayer (d5–d10) stage or both as outlined in Figure 3A. Continuous expression of Pax3 (dox/dox) resulted in MyoD and MyoG expression, and as expected, this was not observed in the noninduced group (cont/cont), as shown by immunofluorescence staining and RT-qPCR for MyoD, as well as Western blot for MyoG (Fig. 3B–3D). Induction of Pax3 from day 5 (cont/dox), only at the monolayer stage, resulted in the expression of MyoD at levels comparable to the group that had been induced from day 2 (Fig. 3B–3D). On the other hand, when Pax3 induction was applied solely at the EB stage (D2–D5) (dox/cont), the Pax3 effect on MyoD expression was much reduced (Fig. 3B–3D). Doxycycline withdrawal after day 10 does not affect expression of MyoD at day 15 (Supporting Information Fig. S2), suggesting that at this point, cells are fully committed to the skeletal myogenic lineage. Along these lines, Myogenin, which is associated with terminal differentiation, is slightly upregulated upon Pax3 removal, both at protein and mRNA level (Supporting Information Fig. S2). These data suggest that the PDGFRα+FLK-1 cell fraction is not fully committed to the myogenic fate after only transient Pax3 exposure (D2–D5), while later withdrawal of Pax3 is associated with terminal differentiation, in agreement with our previous observations [20].

Figure 3.

Temporal windows of Pax3-responsiveness during embryoid body (EB) differentiation. (A): Outline of time windows of Pax3 induction during EB differentiation (D2–D5) and monolayer stage (D5–D10). The following four experimental groups were assessed: (1) no dox added in any differentiation stage (cont/cont), (2) dox added in both stages, EB and monolayer, from day 2 to day 10 (dox/dox), (3) dox added only from day 5 at the monolayer stage (cont/dox), and (4) dox added only at the EB stage, from day 2 to day 5 (dox/cont). (B–D): Pax3 is necessary to drive the myogenic commitment of mesodermal precursors. In all experimental arms, day 5 EBs were sorted for PDGFRα+FLK-1 cells, and then cultured as monolayer, on gelatin-coated glass coverslips for additional 5 days (D5–D10): the first 4 days in the presence of proliferation medium, and in the last 24 hours in differentiation myogenic medium to induce myotube formation, at which point samples were analyzed by immunofluorescence staining (B), real time polymerase chain reaction (PCR) analyses (C) or Western blot (D). Pax3 was induced as outlined in (A). (B): Immunofluorescence staining for MYOD. Cells were fixed and stained with anti-MYOD (red) antibody and DAPI (nuclei, blue). Pax3 induction at the monolayer stage results in expression levels of MYOD that are comparable to continuous induction. Bar = 50 μm. (C): MyoD expression by reverse transcription PCR (RT-qPCR) analysis. Results were normalized to Gapdh. Graphs represent the mean ±SD of at least three independent experiments. *, p < .05; **, p < .01. (D): Western blot analyses for MYOG and PAX3. Samples were lysed with RIPA buffer and analyzed with the indicated antibodies. Histone H3 antibody was used for loading control. Cont/Cont (lane 1), Dox/Dox (lane 2), Cont/Dox (lane 3), and Dox/Cont (lane 4). Abbraviations: DAPI, 4′,6-diamidino-2-phenylindole; dox, doxycycline.

Deletional Analysis of Pax3 Domains

To dissect how the different domains of Pax3 play a role in muscle determination during EB differentiation, we generated different mouse ESC lines which allow the doxycycline-dependent induction of deletion mutants of Pax3 (as shown in Fig. 4A). Besides mutants deleted for the whole paired domain (ΔPD), octapeptide (Δ8), homeodomain (ΔHD), and transactivation domain (ΔTAD), we generated a deletion mutant of the carboxy-terminal part of the paired domain (ΔPD-C), also known as RED domain, and one missing a small conserved region of 40 aminoacids inside the TAD (Δ352–391) (Supporting Information Fig. S3). Following the generation of each respective ESC line, Pax3 induction was confirmed by Western blot using two Pax3-specific antibodies (Fig. 4B). The intracellular localization of the deletion mutants was also assessed by confocal microscopy. To perform this analysis, we generated a further series of cell lines expressing Pax3 and its deletion mutant fused with the 3XFLAG tag. To avoid any interference due to the 3XFLAG sequence, we compared the myogenic differentiation of Pax3-FLAG to the unmodified Pax3, which demonstrated no differences between these two cell lines (data not shown). All the Pax3 deletion mutants localized to the nucleus, in a similar fashion to the full-length (Fig. 4C). Indeed, using nuclear localization signal (NLS) prediction softwares (NucPred and pSortII [39, 40]), the putative NLS represents a short region of six to seven aminoacids N-terminal to homeodomain, which was not altered by any deletion mutant.

Figure 4.

Generation of Pax3 deletion mutants. (A): Schematic representation of Pax3 mutant constructs used to generate inducible embryonic stem cell (ESC) lines. (B): Expression of the Pax3 mutant genes upon Dox induction during embryoid body (EB) differentiation of inducible ESC lines. Day 5 EBs were lysed with RIPA buffer, and protein extracts were analyzed by Western blot using anti-GAPDH and anti-PAX3 antibodies directed toward the amino- (lanes 11–16) and carboxy-terminals (lanes 1–10). Pax3 (lanes 1, 6, 11, and 14); Pax3Δ8 (lanes 2 and 7); Pax3 ΔPD-C (lanes 3 and 8); Pax3 ΔHD-C (lanes 4 and 9); Pax3 ΔPD (lanes 5 and 10); Pax3 ΔTAD (lanes 12 and 15); Pax3 Δ352-391 (lanes 13 and 16). (C): Day 5 EBs were trypsinized, plated on glass coverslips in the presence of dox, and after 2 days, cells were fixed and subjected to immunofluorescence staining with anti-PAX3 antibody. Pictures were acquired with a confocal microscope. Bar = 10 μm. Abbreviations: dox, doxycycline; HD, homeodomain; ΔHD, deletion of HD; PD, paired domain; ΔPD, deletion of PD; ΔPD-C, deletion mutant of the carboxy-terminal part of the paired domain; TAD, transactivation domain; ΔTAD, deletion of transactivation domain; WT, wild type; 8: octapeptide; Δ8, deletion of octapeptide.

The RED Domain Is Important for Pax3-Mediated Paraxial Mesoderm Induction

Since induction of Pax3 in this context increases the frequency of the PDGFRα+FLK-1 cell fraction (Fig. 2A; Supporting Information Fig. S1C and [20]), we therefore investigated the necessity of the different domains of Pax3 during mesoderm differentiation. FACS analysis (Fig. 5A, 5B) shows that the frequency of PDGFRα+FLK-1 cells is unaffected by deletion of the octapeptide (Δ8), suggesting that this sequence, although conserved in some Pax genes, is dispensable for Pax3 paraxial mesoderm function. On the other hand, deletion of the paired domain (ΔPD) and the homeodomain (ΔHD) completely abolished Pax3 activity (Fig. 5A, 5B), confirming the importance of both domains in Pax3-dependent transcriptional activity [41] and the inability of the paired domain only to bind the Pax3 target sequences. Similarly, deletion of the transactivation domain (ΔTAD) resulted in the complete loss of function of Pax3. Among the other mutants tested, deletion of the carboxy-terminal part of the paired domain (ΔPD-C) decreased, but did not abolish, the induction of PDGFRα+FLK-1 cells, indicating that this domain affects the function of Pax3 on paraxial mesoderm. Based on the crystal structure of the paired domain, the carboxy-terminal part is not involved in DNA binding, and it has been proposed to be involved in protein–protein interaction [42]. However, the transactivation domain of Pax3 is not well-characterized and, therefore, the structural basis of Pax3 function is not well-understood.

Figure 5.

Effects of Pax3 mutants on mesoderm patterning. (A): A representative fluorescence-activated cell sorting profile for PDGFRα and FLK-1 in induced and noninduced embryonic stem cells encoding wild-type Pax3 or Pax3 mutated on the selected domains at embryoid body day 5. Transgene expression was induced by adding dox to the culture medium from day 2. Fluorescence intensity for FLK-1 is indicated on the y-axis and PDGFRα on the x-axis. (B): Frequency of paraxial mesoderm progenitors, measured by the percentage of PDGFRα+FLK-1 cells. Data represent mean ± SD of three independent experiments. *, p < .05. Abbreviations: dox, doxycycline; HD, homeodomain; ΔHD, deletion of HD; PD, paired domain; ΔPD, deletion of PD; ΔPD-C, deletion mutant of the carboxy-terminal part of the paired domain; TAD, transactivation domain; ΔTAD, deletion of TAD; 8: octapeptide; Δ8, deletion of 8.

To further investigate the effect of the deletion mutants of Pax3, we evaluated the expression levels of selected markers of paraxial mesoderm or skeletal myogenesis on total day 5 EBs (Fig. 6). In agreement with the FACS data, Pax3 and Pax3 Δ8 have a comparable ability to induce Myf5 and c-Met, two well-characterized targets of Pax3 [6, 7], as well as M-cadherin and Paraxis, confirming the dispensability of the octapeptide. On the other hand, Pax3 ΔPD, ΔHD, and ΔTAD completely lack any ability to induce the paraxial/myogenic markers. Interestingly, deletion of the RED domain (ΔPD-C) drastically impairs the induction of Myf5, M-cadherin, and Paraxis and this is consistent with the reduced ability of this mutant to induce the PDGFRα+FLK-1 cell fraction. However, the ΔPD-C mutant was able to induce c-Met as efficiently as wild-type Pax3, suggesting that this mutation might not completely inactivate Pax3, and that this domain might have a distinct function.

Figure 6.

Gene expression analysis of Pax3 mutants. Relative levels of Myf5, M-cadherin, c-Met, and Paraxis in day 5 embryoid bodies. Transcripts are normalized to Gapdh and reported as relative expression to the noninduced sample. The graphs report the mean ±SD from at least three independent experiments. *, p < .05; **, p < .01. Abbreviations: dox, doxycycline; HD, homeodomain; ΔHD, deletion of HD; PD, paired domain; ΔPD, deletion of PD; ΔPD-C, deletion mutant of the carboxy-terminal part of the paired domain; TAD, transactivation domain; ΔTAD, deletion of TAD; 8: octapeptide; Δ8, deletion of 8.

To determine effects of Pax3 and mutants on other lineages, we analyzed the expression of additional markers to neural crest/melanocyte (Mitf), chondrogenic (Runx2), cardiac (Nkx2-5), and hematopoietic (Runx1) development (Supporting Information Fig. S4). Mitf is a well-known target of Pax3 in the neural crest cells [18] and was induced by wild type (WT), Δ8, and ΔPD-C versions of Pax3 (Supporting Information Fig. S4). Runx2 is a chondrogenic marker expressed in the sclerotome-derived population [43]. Consistent with the general increase of paraxial mesoderm caused by Pax3, Runx2 was induced by the same constructs (Supporting Information Fig. S4). This further supports a role for Pax3 in early somite patterning, as suggested by analysis of splotch mouse embryos [38, 44]. Among the other mesodermal lineages, induction of WT, Δ8, and ΔPD-C versions of Pax3 during EB differentiation resulted in decreased expression of Nkx2-5 (Supporting Information Fig. S4), an early marker of cardiac development [45]. In all cases, the HD, PD, and TAD domain deletions abolished the ability to change gene expression (Supporting Information Fig. S4). In terms of hematopoietic development, no major differences were observed at the levels of Runx1 (Supporting Information Fig. S4).

Myogenic Activity of Pax3 Is Not Compromised by Deletion of the RED Domain

Although the upregulation of Myf5 is indicative of the activation of the myogenic program, irreversible commitment is achieved only by continuous induction of Pax3, ultimately leading to MyoD and MyoG expression (Fig. 3). To determine the ability of these deletion mutants to fully induce the myogenic program, day 5 EBs were plated as monolayers and cultured for 5 days, at which point cells were analyzed by immunofluorescence and Western blot (Fig. 7A). As shown in Figure 7B, Pax3-induced cells were positive for MyoD, while control noninduced cultures did not express this transcription factor. These results were confirmed by Western blot analysis (Fig. 7C). The mutations that abolished the ability to induce the PDGFRα+FLK-1 fraction and the paraxial/myogenic markers (as shown in Figs. 5, 6) also abolished myogenic activity. Interestingly, the ΔPD-C mutation showed a slight decrease in expression of MyoD (Fig. 7B, 7C). As mentioned above, Pax3 induction negatively affects the expression of Nkx2-5 (Supporting Information Fig. S4) and ultimately results in the repression of the cardiac program as shown by immunofluorescence and Western blot for cTnI (Supporting Information Fig. S5). Similar results were reported by Quattrocelli et al. [46]. Notably, the ability of Pax3 to repress the cardiac program inversely correlates with the skeletal myogenic activity of the Pax3 mutants.

Figure 7.

Identification of critical domains for Pax3 activity on the myogenic program. (A): Schematic representation of the experiment layout. To investigate the ability of Pax3 to induce the paraxial mesoderm and the myogenic lineage, unsorted (panel B, C) or sorted PDGFRα+FLK-1 cells (panel D, E) were plated as monolayer on gelatin-coated glass coverslips and cultured for 4 days in proliferation medium and 1 day in differentiation medium. (B): Cells were fixed and stained with anti-MYOD (red) antibody and DAPI (nuclei, blue). Samples were analyzed by epifluorescence microscopy. Bar = 100 μm. (C): Similarly, cells from differentiated cultures were analyzed by Western blot for the expression of MYOD. Cells were lysed with RIPA buffer, and protein extracts were analyzed using anti-MYOD, anti-GAPDH antibodies. Pax3 (lanes 1, 6, 11, and 14); Pax3Δ8 (lanes 2 and 7); Pax3 ΔPD-C (lanes 3 and 8); Pax3 ΔHD-C (lanes 4 and 9); Pax3 ΔPD (lanes 5 and 10); Pax3 ΔTAD (lanes 12 and 15) Pax3 Δ352-391 (lanes 13 and 16). (D): Immunostaining with anti-MYOG antibody (red) and DAPI (nuclei, blue) of differentiated sorted PDGFRα+FLK-1 cells from dox-treated Pax3 and Pax3 ΔPD-C cell lines. Bar = 100 μm. (E): Similar to panel (D), differentiated cultures were analyzed by Western blot with anti-MYOD, anti-MYOG, and anti-PAX3 antibodies. Pax3 (lane 1) and Pax3 ΔPD-C (lane 2). Abbreviations: DAPI, 4′,6-diamidino-2-phenylindole; dox, doxycycline; EB, embryoid body; FACS, fluorescence-activated cell sorting; HD, homeodomain; ΔHD, deletion of HD; PD, paired domain; ΔPD, deletion of PD; ΔPD-C, deletion mutant of the carboxy-terminal part of the paired domain; TAD, transactivation domain; ΔTAD, deletion of TAD; WT, wild type; 8: octapeptide; Δ8, deletion of 8.

We reasoned that mutations specifically interfering with paraxial mesoderm specification but not myogenesis would have such a weak readout in this assay. To test whether the RED domain is required for myogenic commitment, we sorted PDGFRα+FLK-1 cells from both Pax3 and Pax3 ΔPD-C day 5 EBs and cultured sorted presumptive paraxial mesoderm in monolayer. Interestingly, after 5 days of differentiation, we observed that Pax3 bearing the ΔPD-C mutation was able to induce the terminal differentiation at levels comparable to the full length Pax3, as shown by immunostaining for Myogenin and Western blot of MyoD and Myogenin (Fig. 7D, 7E). This indicates that the C-terminal non-DNA-binding part of the paired domain plays an essential role in the early activity on paraxial mesoderm but is dispensable for myogenesis.


In this study, we investigated the role of the transcription factor Pax3 in paraxial mesoderm formation and early myogenesis using differentiating mESCs. Using this system, we show that cells derived from EBs have the potential to differentiate toward the skeletal muscle lineage in a process that depends upon maintained Pax3 expression. This is preceded by upregulation of several somite marker genes, particularly Paraxis and Meox1, which suggests that Pax3 drives the generation of an intermediate population with features similar to the epithelializing somite population. In addition, Pax3 expression resulted in the induction of Myf5, c-Met, Nr2f2, and Prrx1 but not MyoD or Myf6, indicating that Pax3 drives cell fate changes that mimic the events happening in the early myotome. Interestingly, we observed that early markers of paraxial mesoderm (Dll1, Msgn1, and FoxC1) were upregulated in cells representing lateral and cardiac mesodermal fractions (PDGFRαFLK-1+ and PDGFRα+FLK-1+, respectively), while markers associated with the epithelial somite (Meox1 and Tcf15) and myotome formation (Myf5, c-Met, Prrx1 and Nr2f2) were mainly upregulated in the presumptive cardiac and paraxial mesodermal fractions (PDGFRα+FLK-1+ and PDGFRα+FLK-1, respectively). These findings suggest that Pax3 might induce the other mesodermal populations to acquire a paraxial fate and subsequently commit them toward the myogenic lineage. This observation is also in agreement with a previous report showing the ability of Baf60c to induce the cardiac tissue from mouse mesoderm [47], suggesting these cells represent a noncommitted intermediate able to generate different lineages. Although Pax3-induced EBs become primed, maintained expression of Pax3 is required to drive commitment to the myogenic lineage.

Because Pax3 null embryos show defects in both skeletal muscles and vertebrae [38, 44], it has been proposed that Pax3 might be involved in the process of epithelialization of the somite. Using the ES/EB system, we show upregulation of paraxial mesodermal genes following Pax3 induction, with a particular induction of the epithelial somite markers Meox1 and Paraxis, and the myogenic genes Myf5 and c-Met. Notably, our approach identified Dll1 and Meox1 as potential candidate targets of Pax3, even though more analyses are needed to clarify the role of Pax3 in this process. These findings are supported by in vivo studies using homozygous splotch E10.5 embryos, which showed that the expression levels of Paraxis as well as Myf5 and c-Met in the somites were much reduced in the absence of Pax3 [44], while expression of Dll1 was perturbed [38]. Moreover, a Pax3 binding site was reported in the promoter region of the Meox1 gene [16]. Another interesting finding of this study was the upregulation of Nr2f2 and Prrx1. This indicates that in the presence of Pax3, differentiating cells express markers characteristic of the delaminating dermomyotome [36, 37]. Therefore, our data support the involvement of Pax3 in the process of somite differentiation prior to myogenic commitment and provide a cellular model in which to study the molecular mechanism underlying Pax3 function in both the epithelial somite and the dermomyotome.

Our functional analysis using deletion mutants of Pax3 evaluated the requirement for each conserved domain of Pax3 during the induction of both paraxial mesoderm and skeletal muscle differentiation. All the mutants are able to localize to the nucleus, similarly to the full length protein. In silico analyses suggested the presence of a small NLS immediately N-terminal to the homeodomain sequence. Nevertheless, since our deletion strategy excluded some regions of Pax3 (before the PD and between PD and HD), we cannot rule out the possibility that Pax3 localizes to the nucleus upon interaction with a cofactor. Mutants lacking the paired domain or homeodomain were unable to efficiently induce the myogenic program, and these results confirm previous structural studies suggesting a critical role exerted by Pax3 DNA binding domains and the importance of their interaction [29, 41]. Similarly, deletion of the transactivation domain completely abolished Pax3 activity or, as in the case of the Δ352-391 deletion, strongly impaired it. On the other hand, although the octapeptide is conserved in several members of the Pax family of transcription factors and reported to interact with the transcriptional repressor Groucho [14], deletion of this sequence did not affect Pax3 function in any of our assays. This discrepancy could be explained by a different requirement of binding to Groucho between the Pax proteins, since this interaction was reported for the subgroup Pax2/5/8. Our results corroborate the hypothesis raised by Relaix et al. [48], which proposes that Pax3 functions principally as a transcriptional activator during embryogenesis.

Expression of Pax3 during EB differentiation resulted also in the repression of the cardiac program, as evidenced by downregulation of Nkx2-5 and cTnI. This effect correlated with the skeletal myogenic activity of Pax3 since deletion mutants showed a different ability to repress the cardiac program. These findings are corroborated by similar observations published recently by Quattrocelli et al. [46].

Pax3 function in promoting paraxial mesoderm gene expression and induction of PDGFRα+Flk1 cells was impaired by deletion of the C-terminal part of the paired domain, the so-called RED domain. Surprisingly, the RED domain was not required for myogenesis, since PDGFRα+Flk1 presumptive paraxial mesoderm cells expressing Pax3 ΔPD-C were able to differentiate similarly to the full-length Pax3. As shown in the crystal structure of the paired domain [42], and subsequently confirmed by other studies, the RED domain is not in direct contact with DNA. We speculate that this domain interacts with protein cofactors that are specific to paraxial mesoderm formation and not required or absent in later myogenic specification. Identifying the cofactors that interact with this domain may illuminate the context-specific differential effects of Pax3 expression.


Our study shows that using inducible mESCs, we can recapitulate early stages of paraxial mesoderm differentiation, allowing us to study the early function of Pax3. Interestingly, Pax3 upregulates the epithelial somite markers Meox1 and Paraxis, supporting the involvement of this transcription factor in the process of somite epithelialization. Moreover, we conclude that the ability of Pax3 to induce myogenesis is mainly restricted to mesoderm, via the generation of PDGFRα+FLK-1 cells. Time course analysis demonstrated that Pax3 induction is critical to promote the commitment of PDGFRα+FLK-1 cells toward the myogenic lineage, suggesting a distinct function of Pax3 in the induction of paraxial mesoderm versus the myogenic lineage. Finally, our deletion analysis identifies the RED domain (Pax3 ΔPD-C) as an important determinant for Pax3 function in paraxial mesoderm formation, and thus represents a candidate region to identify cofactors specific to early activity of Pax3.


This project was supported by NIH Grants R01 AR055299 and RC1AR058118 to R.C.R.P. We thank Kirsta Hoffman for technical assistance, all the members of the Perlingeiro lab for scientific discussion, and Michael Kyba for critical reading of the manuscript.


The authors declare no potential conflicts of interest.