PAX3-SIX2 muscle stem cell heterogeneity drives muscle-specific repair dynamics diversification Authors: Virginia Zoglio, Sarah Chebouti, Fayez Issa, Théo Massard, Benjamin Polin, Sylvie Manin, Pascal Maire, Dawon Choi, Indigo T.C. Chan, Tom H. Cheung, Joana Esteves de Lima, Frederic Relaix Published: 2026 ABSTRACT The human body contains around 640 distinct muscles, each capable of regeneration following injury through the action of muscle-specific stem cells (MuSCs), that express the transcription factor PAX7. Its paralog, PAX3, a master regulator of embryonic myogenesis, is selectively expressed in a subset of adult quiescent MuSCs. The proportion of PAX3-pos MuSCs varies across muscles. By combining lineage tracing and skeletal muscle injury, we demonstrate that PAX3 drives MuSC diversity and muscle-specific regeneration rates. PAX3-expressing MuSCs display enhanced proliferation and differentiation capacities, enabling a faster response following injury, whereas loss of PAX3 leads to proliferation arrest and cell death. Single-cell RNA-sequencing analyses revealed the specific expression of Six2 in PAX3-pos MuSCs. We show that PAX3 is required for SIX2 expression, and loss of SIX2 in MuSCs reduces proliferation and differentiation rates. Conversely, ectopic Six2 expression promotes proliferation of PAX3-neg MuSCs by directly activating cell cycle pathways. With this work, we establish that PAX3-SIX2 expression correlates with distinct MuSCs behavior, influencing regeneration rates in a muscle-type-dependent context. Our findings highlight a previously unrecognized layer of regulation in MuSC behaviour and muscle repair and suggest that PAX3-SIX2 heterogeneity could be leveraged for targeted therapeutic strategies in muscle-wasting diseases. INTRODUCTION Human skeletal muscles exhibit significant anatomical, functional, and molecular diversity, reflecting their distinct embryonic origins, innervation patterns, and mechanical demands. Skeletal muscle possesses a remarkable regenerative capacity, primarily driven by a resident population of muscle stem cells (MuSCs), also known as satellite cells (1). These cells remain in a quiescent state G0 under homeostatic conditions (2), but rapidly activate, proliferate, and differentiate in response to muscle injury or stress, allowing tissue repair (3–5). Among the key regulators of MuSC are members of the paired box (PAX) family of transcription factors, notably PAX7 and PAX3. While PAX7 is widely recognized as a master regulator of adult MuSC maintenance and self-renewal (5–8), the role of PAX3 in adult muscle stem cells remains less clearly defined. PAX3 is well known for its essential functions during embryonic myogenesis, where it orchestrates the specification, migration, and survival of muscle progenitor cells (9, 10). However, in adult tissues, PAX3 expression is restricted to specific muscle groups and subpopulations of MuSCs (11, 12). Notably, Pax3 is absent in head muscles due to their distinct embryonic origin from non-PAX3-expressing precursors (13). In hindlimbs, minimal PAX3 expression occurs in plantaris, quadriceps, soleus, and gastrocnemius muscles, while moderate levels appear in tibialis anterior (TA) and extensor digitorum longus (EDL) (14). Sacco et al. (15) showed that approximately 12% of hindlimb MuSCs express PAX3 (15). Conversely, PAX3 is highly expressed in the trunk and certain forelimb muscles such as the biceps and diaphragm, suggesting its restriction to a specific subset of muscle stem cells with specialized functions (14, 16). Single-fiber analyses demonstrate non-uniform PAX3 expression patterns across muscle fibers of distinct muscle groups, suggesting dynamic regulation during aging or injury (14). Although PAX3-pos MuSCs tend to be associated with skeletal muscle of certain anatomical locations, the difference of PAX3 expression does not seem to correlate with their embryonic origins, metabolic fiber types, or types of innervating motor neurons. These cells demonstrate unique functional properties, including resistance to irradiation and environmental stress (16, 17). Scaramozza et al. (17) identified a subpopulation of MuSC Mx1+/Pax3+/Pax7+ (isolated via Mx1-Cre lineage tracing) with lower reactive oxygen species (ROS) levels, enhanced survival, and robust clonal expansion under radiation stress, compared to the Mx1−Pax3−Pax7+ counterparts (17). Similarly, der Vartanian et al. (16) demonstrated a dual response to environmental stress among MuSC subpopulations. Exposure to the toxic pollutant (i.e. dioxin) selectively depleted PAX3-neg MuSCs while preserving PAX3-pos MuSCs (16). PAX3-mediated activation of mTORC1 was identified as a critical protective mechanism, akin to the G-alert state (18). However, transplantation studies reveal that the loss of Pax3 impairs engraftment potential and renders these cells more susceptible to stress (16, 17), highlighting the role of Pax3 in modulating MuSC adaptation to stress and environmental challenges. Muscle injury represents another remarkable source of stress (19), resulting in the damage and loss of muscle fibers and activation of MuSC-driven repair (4, 20, 21), mobilizing regenerative and adaptive pathways well beyond the site of damage (18). Despite increasing insights into the intrinsic and extrinsic factors that shape MuSC behavior, the extent to which these differences translate into muscle repair dynamics across distinct muscle groups remains largely unexplored. Recent advances have revealed that MuSCs are not a homogeneous population but instead comprise distinct subtypes with unique molecular signatures and regenerative capacities (22). Heterogeneity within the MuSC compartment arises from differences in their intrinsic self-renewal and commitment dynamics, as revealed by asymmetric division patterns marked by Myf5-driven CRE activity (23). This cellular diversity is further influenced by variations in division rates (24), differential expression levels of the transcription factor PAX7 (25) and the presence of PAX3-expressing subpopulations (10). However, the contribution of PAX3 to this diversity, and its impact on muscle regeneration, represents still a subject of active investigation. Together, these observations raise the possibility that PAX3 expression defines a functionally specialized subset of adult MuSCs. However, whether PAX3-positive (pos) and PAX3-negative (neg) MuSCs play distinct roles during tissue repair remains unclear. In particular, it is unknown whether the heterogeneous distribution of PAX3-expressing MuSCs across muscles contributes to differences in regenerative dynamics between muscle groups, and what molecular mechanisms might underlie such specialization. Here, we hypothesized that PAX3 defines a functionally distinct MuSC state that confers a distinct regenerative response and contributes to muscle-specific repair dynamics. To test this hypothesis, we combined lineage tracing, muscle injury models, and single-cell transcriptomics to investigate how PAX3-expressing MuSCs behave during regeneration and to identify molecular pathways underlying their functional specialization. We uncovered that a PAX3-SIX2 axis underlies distinct MuSC behaviors that drive muscle-specific regeneration rates. We revealed that PAX3-SIX2 expression promotes rapid tissue repair through increased proliferation, differentiation, and fusion of MuSCs, whereas PAX3-neg MuSCs prioritize self-renewal to maintain the stem cell pool. Crucially, PAX3 is essential for preserving MuSC survival and timely regeneration in muscles like the biceps brachii, with its loss resulting in delayed repair. These findings uncover a previously unrecognized regulatory layer of MuSC heterogeneity that shapes inter-muscle differences in regenerative kinetics, providing a mechanistic foundation to leverage PAX3-SIX2 heterogeneity for targeted therapies in muscle-wasting diseases. RESULTS Muscles enriched in PAX3-positive MuSC display accelerated regeneration dynamics If PAX3-pos MuSCs represent a functionally distinct stem cell state, their uneven distribution across skeletal muscles could translate into differences in regenerative behavior. We therefore sought to determine whether muscles enriched in PAX3-expressing MuSCs exhibit distinct regeneration dynamics following injury. Although heterogeneity in Pax3 expression among MuSCs has previously been reported using transcriptional reporters or transcriptomic analyses, it remains unclear whether this variability translates into measurable differences in PAX3 protein abundance across skeletal muscles. To address this question, we quantified PAX3 protein levels by Western blot in the muscle groups used for our injury assays. We observed heterogeneous PAX3 levels across distinct muscles groups, including diaphragm (DIA), biceps brachii (BCPS), rectus femoris (RF), tibialis anterior (TA) and gastrocnemius (GASTROC) muscles (Fig. 1, A and A’), in line with previous reports (14, 16). The Tomato lineage tracing of PAX3-derived MuSCs and myofibers progeny in the adult (4-week-old) Pax3CreERT2/+;R26stopTomato/+ mice, following tamoxifen induction, confirmed variable contributions of PAX3-pos MuSCs to muscle fibers across anatomical regions (Fig. 1B). This regional heterogeneity was recapitulated at the transcript level. Pax3 mRNA was significantly enriched in MuSCs FACS-isolated from forelimb and trunk muscles compared to hindlimb-derived MuSCs (Fig. 1C), whereas Pax7 transcript levels were comparable between muscle compartments (Fig. 1D). Quantitatively, Pax3 expression was largely absent or restricted to a minor fraction (∼10%) of MuSCs in hindlimb muscles such as the GASTROC, with intermediate representation (∼20%) in the TA. In contrast, a substantial proportion (∼50%) of MuSCs in trunk and forelimb muscles, including BCPS and diaphragm, retained robust Pax3 expression (Fig. 1E). Surprisingly, a substantial proportion of PAX3-pos MuSCs (∼35%) is observed in the rectus femoris muscle (RF) (Fig. 1E), in accordance with our previous observation that embryonic origin or rostro-caudal position does not determine the spatial distribution of PAX3-pos MuSCs across muscles (10). Based on these observations, hereafter we classified TA and GASTROC as PAX3-low muscles and BCPS and RF as PAX3-high muscles. The PAX3-pos subpopulation is associated with enhanced resistance to stress (17, 18), supporting the notion that the spatial distribution of PAX3-pos MuSCs could underlie muscle-specific differences in regenerative capacity. PAX3-positive MuSC distribution across muscles correlates with regeneration efficiency. (A) Western blot analysis of PAX3 and H3 in diaphragm (DIA), biceps brachii (BCPS), rectus femoris (RF), tibialis anterior (TA), and gastrocnemius (GASTROC) muscles (n = 4). (A’) Quantification of PAX3 signal intensity from three independent blots. (B) Tomato lineage tracing in Pax3CreERT2/+;R26stopTomato/+ muscles. (C and D) RT-qPCR for Pax3 and Pax7 in FACS-isolated MuSCs from forelimb/trunk and hindlimb muscles (n = 6). (E) Quantification of PAX7-pos PAX3-pos (β-gal -pos) cells (blue) and PAX7-pos PAX3-neg (β-gal-neg) cells (orange) in resting muscles. (F to I) Immunofluorescence staining and quantification of eMHC, MYOG, LAMININ, DYSTROPHIN, and DAPI in injured muscles at indicated time points. (J) Experimental timeline for intramuscular BaCl2 injection, created in BioRender. Zoglio, V. (2026) https://BioRender.com/lau6rvy. (K) Representative staining of PAX7, DYSTROPHIN, b-gal, and DAPI in BCPS muscle. Yellow arrows indicate PAX7-pos PAX3-pos (β-gal-pos) MuSCs, whereas arrowheads highlight the PAX7-pos PAX3-neg (β-gal-neg) population. (L and M) Quantification of PAX7-pos cells in BCPS and TA muscles at indicated times after injury. Data are mean ± SD or SEM as indicated; Mann-Whitney test. NS, not significant. Scale bar, 40 μm. To determine whether spatial differences in PAX3 MuSC translate into functional disparities in regeneration, we performed BaCl2-mediated injury in muscles with distinct percentages of PAX3-pos MuSCs. (Fig. 1F). Strikingly, the repair dynamics assessed using embryonic MHC staining (MYH3), which is transitorily expressed in newly formed fibers during regeneration, varied between different muscle groups (Fig. 1, F and G). Although all muscles displayed robust MYH3-pos fibers at 5 days post-injury (dpi), regenerative progression diverged thereafter (Fig. 1, F and G). PAX3-low muscles exhibited a higher number of MYH3-pos fibers at both 7 and 14 dpi compared with PAX3-high muscles (Fig. 1, F and G). Importantly, the persistence of MYH3-pos fibers inversely correlated with the proportion of PAX3-pos MuSCs present in each muscle, consistent with faster regenerative progression in PAX3-enriched muscles (fig. S1D). Consistently, PAX3-low muscles presented fibers with smaller cross-sectional area (CSA) compared to PAX3-high muscles (fig. S1, A to C). Together, the faster decrease in MYH3-pos fibers and increased CSA suggest an accelerated fiber maturation potential in PAX3-pos muscles. To link these observations with MuSC differentiation dynamics in each muscle type, we quantified the number of MYOG-pos cells at 4-, 7- and 14dpi. PAX3-high muscles exhibited a more pronounced decline in MYOG-pos cells over time, consistent with faster progression of myogenic differentiation and myofiber maturation (Fig. 1, H and I). In line with the differences in regeneration kinetics, PAX3-low muscles displayed delayed resolution of fibro-adipogenic progenitors reflected by a higher percentage of PDGFRα-pos cells at 14dpi (fig. S1, E and F). In addition, PAX3-low muscles exhibited prolonged macrophage infiltration, with increased numbers of F4/80-pos cells at 7- and 14dpi compared with PAX3-high muscles (fig. S1, G and H). Together, these observations demonstrate that these skeletal muscles differ in their regenerative kinetics and maturation efficiency. We propose that the observed functional disparities are driven by intrinsic, PAX3-dependent specialization of resident MuSC populations, which shapes muscle-specific regenerative competence. To mechanistically link PAX3 expression to MuSC function, we performed longitudinal β-galactosidase lineage tracing on a Pax3nlacZ-IRESGFP/+ knock-in mouse model (16), enabling precise monitoring of PAX3-pos MuSCs dynamics during regeneration. The injured and the contralateral TA and BCPS muscles were harvested 2-, 4-, 7- and 28dpi to perform histological assessments (Fig. 1J). MuSCs kinetics were established by performing quantification of the total PAX7-pos population, which reaches a peak at 4dpi in both injured muscles (Fig. 1, K to M). Interestingly, while the number of MuSCs at homeostasis was the same between both muscles, in the PAX3-high BCPS muscle (Fig. 1L), the total number of MuSCs at 4dpi was double compared to the PAX3-low TA muscle (Fig. 1M), suggesting that the early repair is driven by faster amplification or accelerated differentiation of the MuSC population in the PAX3-high muscle. Using the Pax3 β-gal reporter, we quantified the proportion of PAX3-pos and PAX3-neg MuSCs within the total PAX7 population in both injured muscles. While BCPS and TA muscles show distinct levels of PAX3-pos (β-gal-pos) MuSCs in homeostasis, we observed a striking shift in their relative proportions during regeneration: the PAX3-pos fraction decreased transiently at 2-, 4-, and 7dpi in both muscles, with restoration by 28 dpi (Fig. 1, L and M). This result highlights the dynamic shift in balance between PAX3-pos and PAX3-neg subpopulations during repair, suggesting that either the PAX3-neg population exhibits increased proliferation during tissue repair, or that the PAX3-pos population undergoes faster differentiation, resulting in a transient accumulation of PAX3-neg cells, which may subsequently contribute more substantially to the self-renewing pool. PAX3-driven MuSCs heterogeneity controls MuSCs specialized function in response to tissue damage To investigate the biological processes underlying the decrease in PAX3-pos MuSCs, we examined the self-renewal, commitment and differentiation capacity of MuSC sub-populations. First, we investigated the quiescent state of MuSCs in homeostasis by using the quiescence marker Calcitonin receptor (CALCR) (26). We observed that the number of CALCR-pos MuSCs was comparable between PAX3-pos and PAX3-neg MuSCs within resting BCPS and TA muscles (fig. S2, A to D). This indicates that both subpopulations maintain comparable quiescence patterns under steady-state conditions, while exhibiting divergent responses following injury (Fig. 1, L and M), suggesting a bimodal regulatory mechanism dependent on intrinsic differences in their myogenic potential and regenerative capacity. During regeneration, strikingly, the percentage of MuSCs double positive for PAX3 and MYOD (marker for committed MuSCs) was significantly higher compared to the PAX3-neg MYOD-pos MuSCs, at 4- and 7dpi (fig. S2, E, F, and G), suggesting a higher commitment of the PAX3-pos MuSCs towards the myogenic lineage. Co-immunostaining with the differentiation marker MYOG and the PAX3 reporter GFP revealed that PAX3-derived MuSCs (GFP-pos) display a greater ability to terminally differentiate in response to injury (Fig. 2, A and C), while PAX3-neg cells mostly retained a quiescent profile (CALCR-pos) (Fig. 2, B and D). Our results revealed a functional heterogeneity linked to Pax3 expression, as indicated by the distinct contribution to differentiation (PAX3-pos MuSC) or self-renewal (PAX3-neg MuSC) (Fig. 2, A to D). This bimodal behaviour arises during muscle repair, with PAX3-derived MuSCs preferentially contributing to regeneration, while PAX3-neg cells preferentially replenish the quiescent pool, ensuring long-term stem cell maintenance. PAX3 confers functional heterogeneity in MuSC response to muscle injury and stress cues, with higher engrafting potential. (A) Representative co-immunofluorescence staining for GFP (PAX3 reporter), PAX7, MYOGENIN, and DAPI in 4 dpi and uninjured BCPS muscle sections. (B) Representative co-immunofluorescence staining for GFP, PAX7, CALCR, and DAPI in 7 dpi and contralateral uninjured BCPS muscle sections. (C and D) Quantification of PAX3-pos and PAX3-neg differentiating MYOG-pos cells and CALCR-pos self-renewing cells, normalized to the corresponding total PAX3-pos or PAX3-neg populations. (E) Experimental design quantification of EdU incorporation in vivo. Created in BioRender. Zoglio, V. (2026) https://BioRender.com/tptl9vw. (F) Quantification of (E) showing the percentage of PAX3-pos GFP-pos EdU-pos cells (in blue) and PAX3-negGFP-neg EdU-pos myoblasts (in orange) (n = 6 independent experiments). (G and H) In vitro proliferation assay showing EdU pulse labeling in GFP-pos and GFP-neg myoblasts, with quantification of EdU-pos PAX7-pos cells normalized to DAPI-pos nuclei. (I) Experimental time-line of transplantation experiment. Created in BioRender. Zoglio, V. (2026) https://BioRender.com/exbxo45. (J to M) Representative TA sections, and quantification of TOMATO-pos fibers and TOMATO-pos PAX7-pos cells 6 days post-engraftment. Scale bar, 200 μm. (N to P) FACS-based culture strategy created in BioRender. Zoglio, V. (2026) https://BioRender.com/kxoaqxo; representative immunostaining, and quantification of GFP-posPAX7-pos, GFP-negPAX7-pos, and PAX7-pos myoblasts after 72 h in culture. (n = 3 independent proliferation assays). Error bars, mean ± SD, Mann-Whitney test. Scale bar, 40 μm. The finding that MuSCs derived from the PAX3 lineage are associated with increased differentiation in vivo, led us to hypothesize that PAX3-derived MuSCs can potentially provide a faster activation kinetic in response to injury. To test this hypothesis, we performed in vivo EdU labeling, immediately after injury, in BCPS and RF muscles from Pax3nlacZ-IRESGFP/+ mice (Fig. 2, E and F). Injured muscles were injected 42-hours post-injury with the nucleotide analog 5-ethynyl-2′-deoxyuridine (EdU) to label cells entering their first S-phase, followed by FACS isolation of GFP-pos (PAX3-derived) and GFP-neg MuSCs. Remarkably, 42-hours after injury, ∼35% of GFP-pos MuSCs were EdU-pos compared to ∼17% of GFP-neg MuSCs, demonstrating that PAX3-lineage MuSCs break quiescence and enter cell cycle significantly more rapidly after injury (Fig. 2, E and F). These findings directly confirm the enhanced activation kinetics of PAX3-pos MuSCs during the early phases of tissue repair. Consistently, in vitro EdU incorporation revealed a higher proliferative capacity of GFP-pos cells (Fig. 2, G and H). To evaluate in vivo the fusion and repopulation potential, PAX3-pos and PAX3-neg cell preparations were transplanted into the TA muscles of Rag2−/−: γC−/− host mice (25) (Fig. 2I). Prior to cell transplantation, both TAs were subjected to BaCl2 injection, to establish a permissive regenerative environment. On the day of transplantation, MuSCs were isolated from donor mice by FACS based on TOMATO (TOM) reporter expression using either Pax3CreERT2/+;R26stopTomato/+ mice (27) or Pax7CreERT2/+;R26RmTmG/+ mice (28) (Fig. 2I). Specifically, TOM-pos MuSCs obtained from the forelimb and trunk muscles of Pax3CreERT2/+;R26stopTomato/+ mice marked the PAX3-derived lineage. In parallel, TOM-pos MuSCs isolated from the hindlimb muscles (which contain a minority of PAX3-pos MuSCs) of Pax7CreERT2/+;R26RmTmG/+ mice served as the PAX3-neg control population. Following isolation, donor MuSCs were selectively transplanted into the right and left TA muscles of each host mouse, enabling direct within-animal comparison of the two cell populations. Six days after engraftment, the grafted muscles were harvested for subsequent analysis. Remarkably, PAX3-pos/TOM-pos MuSCs from donors Pax3CreERT2/+;R26stopTomato/+ exhibited significantly enhanced engraftment and contributed robustly to new myofiber formation, as evidenced by a greater number of donor-derived fibers within the host muscle (Fig. 2, J and K). Conversely, PAX7-pos/TOM-pos MuSCs were predominantly retained within the stem cell compartment, as shown by the PAX7-pos cell number quantification and under basal lamina localization (Fig. 2, L and M). Together, these findings reveal that PAX3-derived MuSCs constitute a subpopulation with primed higher regenerative capacity, characterized by faster activation, proliferation, and enhanced differentiation following injury, while PAX3-neg cells predominantly sustain the quiescent stem cell pool. To further investigate whether these intrinsic differences persist outside the MuSCs native niche and considering that in vitro culture acts as a physiological stressor, we assessed the behavior of GFP-pos and GFP-neg populations with cell culture analysis on FACS-isolated MuSCs from forelimb muscle groups (Fig. 2N). First, GFP-pos (PAX3-pos), GFP-neg (PAX3-neg) and a mixed population (PAX7-pos) were cultured in low density and high-serum conditions, representing a proliferative environment. Strikingly, 72 h post-plating we observed that the GFP-pos cells display a 3-fold expansion in the number of nuclei per field compared to GFP-neg cells in culture (Fig. 2, O and P). Consistently with the higher proliferative signature, when cultured in low-serum conditions GFP-pos myoblasts displayed higher differentiation capacity, as observed by MYOG (fig. S2, H and I) and myosin staining (MF20 antibody) (fig. S2, J and K). To exclude that the higher density in the proliferation assay would differentially affect myoblast fusion abilities in the two conditions, we cultured an equally distributed mix of freshly isolated myoblasts (GFP-pos and GFP-neg) after FACS and quantified the number of nuclei in multinucleated myotubes. The fusion index result showed that the percentage of myonuclei that inherited β-gal from PAX3-pos cells (β-gal-pos/MF20-pos) is higher compared to the counterparts (β-gal-neg/MF20-pos), thus confirming the stronger ability of GFP-derived myoblasts to differentiate and fuse (fig. S2, L and M). Notably, GFP-neg myoblasts failed to generate β-gal-pos cells after clonal expansion (fig. S2, N and O), indicating that PAX3-pos and PAX3-neg cells represent stable, lineage-restricted states. On the contrary, the stability of the GFP reporter precluded tracing of PAX3-derived myoblasts in vitro, requiring live-imaging approaches for future studies. These findings establish PAX3-derived myoblasts as a primed subpopulation, with enhanced response to injury and culture-induced stress, enabling rapid activation, proliferation, differentiation and fusion. PAX3 is required for survival of PAX3-expressing MuSCs To explore the functional relevance of Pax3 expression in MuSCs in vivo, we conditionally deleted Pax3 in MuSCs by intercrossing the Pax3CreERT2/+ driver (28), with Pax3flox/+ (29) and R26stop-Tomato (30) reporter mice (Pax3cKO mice). Tamoxifen-induced CRE recombination enabled deletion of Pax3 in MuSCs of 3-month-old mice, with Tomato expression marking PAX3-derived cells (Fig. 3, A and B). Cre expression was induced by feeding the mice for two weeks with a tamoxifen diet (Fig. 3A). The downregulation of Pax3 was validated by RT-qPCR, showing a loss of Pax3 mRNA expression of 98% (Fig. 3, B and C), whereas Pax7 expression was not altered (Fig. 3D). Next, we assessed whether the conditional deletion of Pax3 would affect resting BCPS and TA muscles. Quantification of PAX7-pos MuSCs showed no difference in MuSC numbers in homeostasis (fig. S3, A to C). Similarly, no significant disruption in the proportion of PAX3-pos versus PAX3-neg MuSCs is observed in resting BCPS and TA muscles upon Pax3 deletion (fig. S3, D and E). However, following injury, BCPS muscles carrying a MuSC-specific conditional deletion of Pax3 exhibited a significant reduction in the total PAX7-pos MuSC pool at 7 dpi (Fig. 3E), whereas TA muscles were not affected (fig. S3, F to H). Pax3 is required for survival and maintenance of PAX3-positive MuSCs in BCPS muscle. (A) Experimental timeline for tamoxifen treatment, BaCl2 injury, and muscle collection created in BioRender. Zoglio, V. (2026) https://BioRender.com/m6pqdrg. (B) Recombination efficiency assessed by Pax3 mRNA reduction after CRE activation. (C and D) RT-qPCR analysis of Pax3 and Pax7 in PAX3-pos, PAX3-neg, and Pax3cKO MuSCs. (E) Quantification of total PAX7-pos MuSCs after Pax3 inactivation in 7 dpi BCPS muscle. (F to I) Representative co-immunofluorescence staining using antibodies against TOMATO (PAX3 reporter, red), PAX7 (green), DYSTROPHIN (grey), and nuclei (DAPI, blue) on BCPS muscle sections collected at 2 dpi (F) and 7dpi (H) from the control and the Pax3cKO mouse model. Yellow arrows indicate PAX3-pos MuSCs, whereas arrowheads highlight PAX3-neg population. (J) Representative co-immunofluorescence staining using antibodies against TOMATO (PAX3 reporter, red), PAX7 (green), EdU (white), and nuclei (DAPI, blue) on FACS-isolated myoblasts. (K) Quantification of (J) showing the percentage of EdU-pos TOM-pos PAX7-pos (red) versus EdU-pos TOM-neg PAX7-pos MuSCs (green) and EdU-pos TOM-pos (Pax3cKO) PAX7-pos (dark red), normalized to total number of nuclei. (L) Representative co-immunofluorescence staining using antibodies against TOMATO (PAX3 reporter, red), Tunel (green) and nuclei (DAPI, blue) on FACS-isolated myoblasts derived from control and Pax3cKO mouse model. (M) Quantification of (L) showing the percentage of Tunel-pos TOM-pos (red) versus Tunel-pos TOM-neg MuSCs (green) and Tunel-pos TOM-pos (Pax3cKO) (dark red), normalized to total number of nuclei. (N) TUNEL quantification in 2 dpi BCPS sections. (O and P) γH2AX staining and quantification in FACS-isolated myoblasts from control and Pax3cKO mice. Scale bars, 10 μm. Data are mean ± SD or SEM as indicated; Mann-Whitney test. NS, not significant. Scale bars, 40 μm. Remarkably, in BCPS muscles, the PAX7-pos MuSC pool was significantly reduced at all time points following injury in the Pax3cKO model compared to controls, reflecting the loss of the PAX3-pos (TOM-pos) subpopulation (Fig. 3, F and G). This was further supported by the striking absence of Tomato lineage labeling in Pax3cKO muscles, whereas in control muscles PAX3-pos TOM-pos MuSCs actively contributed to regeneration by generating TOM-pos myofibers (Fig. 3, F, H, and I). These findings demonstrate that PAX3 is essential for maintaining the PAX3 MuSC pool during BCPS muscle repair. To evaluate the fate of Pax3cKO MuSCs, we analysed cell survival and proliferation rate in culture (Fig. 3, J and K). FACS-isolated Pax3cKO myoblasts showed reduced proliferation, seen by EdU incorporation (Fig. 3, J and K), increased apoptosis (TUNEL-pos cells) (Fig. 3, L and M) and higher DNA damage (γH2AX levels) (Fig. 3, O and P), linked to a mitotic catastrophe phenotype (31). Notably, in vivo TUNEL assays revealed PAX3-pos cell loss as early as 2dpi (Fig. 3N), underscoring the role of Pax3 as a bona-fide survival factor for MuSCs in PAX3-enriched muscles. Histological analysis further revealed that deletion of Pax3 in PAX3-pos MuSCs impaired the regeneration of BCPS muscles. Regeneration was assessed by quantifying MYH3-pos fibers. Pax3cKO BCPS at 7dpi exhibited a higher proportion of MYH3-pos fibers (fig. S3, I and J). While this proportion strongly declined by 28 dpi, MYH3-pos fibers remained slightly increased in Pax3cKO BCPS (fig. S3K). In addition, a reduction in myofiber cross-sectional area (CSA) was observed in 7dpi BCPS muscles of Pax3cKO mice (fig. S3, L and M), but not in TA muscles (fig. S3N), indicating that loss of Pax3 delayed muscle regeneration in BCPS. These results indicate that PAX3 is essential for the maintenance, survival and genomic stability of PAX3-pos MuSC during regeneration, with its loss leading to delayed repair in muscles enriched in PAX3-pos MuSCs. SIX2 acts downstream of PAX3 in adult myogenesis Given the distinct role and function of PAX3-pos MuSCs, we hypothesized that this MuSC subpopulation may present a specific transcriptomic signature, potentially depending on Pax3 expression. To test this hypothesis, we performed single cell RNA-sequencing analysis on FACS-isolated and activated MuSCs from 5dpi BCPS and RF muscles (Fig. 4, A and B). Our analysis confirmed a distinct transcriptomic profile between the PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs (fig. S4, A and B, table S3). The integrated UMAP projection revealed a core myogenic cluster, maintained in both PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) subpopulations, characterized by the expression of key myogenic genes that delineate the progression from quiescence to differentiation (i.e. Calcr, Pax7, Pax3, Myod1, Myf5, Myog) (Fig. 4, A and B). Non-myogenic clusters (IC, SMC, FAP, and ENDO) were excluded from further analysis, likely reflecting minor FACS cell contaminations (fig. S4A). Pax3 and Pax7 transcripts display distinct expression profiles during myogenic progression (Fig. 4, C and D). While Pax7 is gradually downregulated during myogenic progression, Pax3 is rapidly decreased but shows significant upregulation in activated MuSCs and myoblast clusters (Fig. 4D). PAX3-SIX2 promotes the pro-regenerative capacity of PAX3-positive MuSCs. (A and B) scRNA-seq analysis of PAX3-pos and PAX3-neg MuSCs isolated from trunk and forelimb muscles at 5 dpi, showing UMAP projection of the myogenic cluster and expression of myogenic markers, including Pax7, Pax3, Myf5, Myod1, Myogenin and Calcr. (C) Violin-plot showing the distinct level of Pax7 transcript expression between PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs in each myogenic cluster. (D) Violin-plot showing the distinct level of Pax3 transcript expression between PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs in each myogenic cluster. (E) Violin-plot showing the distinct level of Six2 transcript expression between PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs in each myogenic cluster. (F) Validation of sc-RNAseq analysis showing the Six2 mRNA relative level in PAX3-pos (GFP-pos), PAX3-neg (GFP-neg) and Pax3cKO MuSCs by RT-qPCR. (G and H) Co-immunostaining and quantification of myosin-positive nuclei after Six2 knockdown (KD) in GFP-pos and GFP-neg myoblasts. (I and J) Co-immunostaining and quantification after Six2 overexpression (OE) in GFP-pos and GFP-neg myoblasts. (K) Pathway enrichment analysis of genes differentially expressed after Six2 overexpression. (L) Heatmap of cell-cycle genes in Six2-overexpressing PAX3-pos and PAX3-neg MuSCs. (M) RT-qPCR validation of Six2 expression in MuSCs from forelimb, trunk, and hindlimb muscles. Mean ± SEM. Mann-Whitney test. Scale bars, 40 μm. We compared the transcriptomic signature between PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs. PAX3-pos (GFP-pos) displayed significant upregulation of 258 genes, including signaling pathways such as Bmp or Notch signaling, and transcription factors such as Six2 genes, among others, whereas the PAX3-neg (GFP-neg) population were further characterized by increased expression of 109 genes, including transcription factors such as Sox4 or Sox9 genes (fig. S4B and table S3). Consistently, Gene Ontology analysis shows that pathways enriched in PAX3-neg (GFP-neg) MuSCs are associated with metabolic processes such as oxidative phosphorylation and mitochondrial or cytoplasmic translation. In contrast, PAX3-pos (GFP-pos) MuSCs are enriched for terms related to the negative regulation of apoptosis, enhanced IL-6 production, and modulation of BMP signaling, supporting our findings that PAX3-pos MuSCs exhibit enhanced activation, proliferation, and rapid tissue repair (fig. S4, C and D). Given previous evidence that PAX3 cooperates with SIX family transcription factors to regulate developmental myogenesis (32), we decided to focus on Six2 which is significantly expressed in the PAX3-pos activated and myoblast clusters (Fig. 4E). This specificity of Six2 to the PAX3-pos population was further confirmed by RT-qPCR on FACS-isolated MuSCs (Fig. 4F). Consistently, Six2 expression was lost in Pax3cKO MuSCs (Fig. 4F), suggesting a genetic cross-talk between PAX3 and SIX2. Surprisingly, while Six2 mRNA was restricted to PAX3-pos cells, the other members of the SIX protein family did not show any specificity (fig. S4, E to G). SIX2 is part of the homeoprotein family of transcription factors that is known to initiate the myogenic program in the embryo through the activation of Myf5/Myod1 (32). We therefore hypothesized that a regulatory PAX3-SIX2 axis could be linked to the specific functional behaviour of PAX3-pos MuSCs. To test this hypothesis, we performed in vitro gain- and loss-of-function studies by transfecting FACS-isolated MuSCs with a non-targeting (control) or Six2 siRNA (Six2 knock-down KD, Fig. 4, G and H, and fig. S4, H and K); and with an empty plasmid or a plasmid expressing Six2 (Six2 overexpression OE, Fig. 4, I and J, and fig. S4, L and N). Following confirmation of Six2 knock-down (fig. S4I) and up-regulation (fig. S4M), we performed in vitro proliferation and differentiation assays. These experiments demonstrated that PAX3-pos(GFP-pos) MuSCs showed reduced differentiation upon Six2 KD (Fig. 4, G and H, and fig. S4, H and K), similar to PAX3-neg (GFP-neg) control cells, whereas PAX3-neg(GFP-neg) cells displayed an enhanced proliferation capacity upon Six2 OE (Fig. 4, I and J, and fig. S4, L and N), mimicking PAX3-pos(GFP-pos) control cell behavior. Notably, Six2 KD did not impact Pax7 transcript levels (fig. S4J). These results support SIX2 acting as a downstream effector of PAX3 function in MuSCs. To investigate the molecular basis of the PAX3-SIX2 heterogeneity, we performed RNA-seq analysis on PAX3-pos and PAX3-neg MuSCs following Six2 overexpression in vitro. Volcano plot analyses comparing control and Six2-overexpressing (OE) cells revealed 1058 and 1226 upregulated genes in the PAX3-pos and PAX3-neg populations, respectively. Meanwhile, 513 and 902 genes were downregulated in PAX3-pos and PAX3-neg MuSCs, respectively (fig. S5, A and B). Pathway enrichment analysis of PAX3-pos and PAX3-neg MuSCs transfected with Six2 plasmid compared to their respective controls, showed that Six2 OE activated multiple cell cycle-related pathways and markedly increased the expression of proliferation-associated genes (Fig. 4, K and L) in both populations, indicating that SIX2 promotes a proliferative, early myogenic state that enhances MuSC expansion. Our results demonstrate that PAX3-SIX2 heterogeneity underlies the functional specialization of MuSCs during tissue repair. Consistently, Six2 transcripts were more abundant in forelimb and trunk muscles compared with whole hindlimb muscles (Fig. 4M), mirroring the spatial pattern of Pax3 expression (Fig. 1C). To further dissect the functional relationship between PAX3 and SIX2, we tested whether SIX2 could compensate for Pax3 loss in the KO model. However, the rapid depletion of TOM-pos cells post-injury (Fig. 3, F and G) and in culture (Fig. 3, L and M) required an alternative strategy. Therefore, we employed a tamoxifen-inducible system to achieve acute Pax3 deletion in vitro by treating FACS-isolated MuSCs with 4-OHT. Recombination efficiency was confirmed by Tomato fluorescence and RT-qPCR, which showed robust Pax3 downregulation within 48 hours (Fig. 5A and fig. S5, C, D, and E). Remarkably, transfection with a Six2-expressing plasmid fully rescued the proliferation deficit (Fig. 5, B and C), prevented cell death (Fig. 5, D and E), and DNA damage (Fig. 5, F and G) in Pax3-deficient myoblasts. To further investigate the mechanistic basis by which Six2 restores viability and reduces DNA damage in Pax3-deficient MuSCs (Fig. 5, E and G), we performed an in-depth analysis of the Six2 OE RNA-seq data specifically in PAX3-neg MuSCs. Reactome pathway enrichment analyses revealed that Six2 OE in these cells drives a remarkably broad and coordinated reactivation of the cell cycle transcriptional program, with enrichment scores reaching −log10(p-value) of up to 40 for Cell Cycle, ∼30 for Cell Cycle Mitotic, and ∼23 for Cell Cycle Checkpoints pathways (fig. S5F). This signature marked all major cell cycle phases, including G1/S transition, S phase, G2/M checkpoints, and mitotic execution (fig. S5, H to K) and was confirmed at the gene level by consistent upregulation of canonical effectors including Ccnb1/2, Ccna2, Cdk1/6, Plk1, Aurka/b, Mcm3–6, Bub1b, Chek1/2, and Mad2l1 across multiple gene sets (fig. S5, H to K). Critically, Six2 OE in PAX3-neg MuSCs concurrently enriched a suite of replication-coupled DNA repair pathways, including Chromosome Maintenance, HDR through Homologous Recombination (HRR), Homology Directed Repair, Resolution of D-loop Structures, DNA Double-Strand Break Repair, and Regulation of TP53 Activity (fig. S5G). Together, these findings support a model in which Six2 functions as a transcriptional activator of the cell cycle in MuSCs, and its overexpression in Pax3-deficient cells restores viability by enabling repair-coupled cell cycle re-entry, thereby resolving the DNA damage accumulation that characterises the Pax3-deficient state (Fig. 5, E and G). SIX2 fully rescues the proliferation deficit, preventing cell death and mitotic catastrophe in Pax3-deficient myoblasts. (A) Experimental time-line indicating in vitro tamoxifen administration strategy and Six2 plasmid transfection on Pax3cKO myoblasts created in BioRender. Zoglio, V. (2026) https://BioRender.com/lau6rvy. (B) Representative co-immunofluorescence staining using antibodies against EdU (white), and nuclei marker (DAPI, blue) on FACS-isolated myoblasts derived from control and Pax3cKO mouse model upon control and Six2 plasmid transfection. Scale bar, 40 μm. (C) Quantification of (B) showing the percentage of EdU-pos cells normalized on total number of nuclei. Means ± SEM (n = 6), P value calculated by Mann-Whitney test. NS, not significant. (D) Representative co-immunofluorescence staining using antibodies against Tunel (green) and nuclei marker (DAPI, blue) on FACS-isolated myoblasts derived from control and Pax3cKO mouse model upon control and Six2 plasmid transfection (Six2 OE). Scale bar, 40 μm. (E) Quantification of (D) showing the percentage of Tunel-pos cells normalized on total number of nuclei. Means ± SEM (n = 4/6), P value calculated by Mann-Whitney test. NS, not significant. (F) Representative co-immunofluorescence staining using antibodies against γH2AX (green) and nuclei marker (DAPI, blue) on FACS-isolated myoblasts derived from control and Pax3cKO mouse model upon control and Six2 plasmid transfection. Scale bar, 10 μm. (G) Quantification of (F) showing the percentage of γH2AX-pos foci per nucleus. Means ± SEM (n = 3), P value calculated by Mann-Whitney test. NS, not significant. This functional rescue underscores the pivotal role of SIX2 as an indirect downstream effector of PAX3 and highlights a previously unrecognized capacity for targeted molecular intervention to restore MuSCs function in the context of PAX3 deficiency. DISCUSSION Our study redefines MuSC heterogeneity by establishing PAX3-pos MuSCs as a functionally distinct, primed subpopulation, essential for an adaptive response to muscle injury. Remarkably, we demonstrate that spatial distribution of PAX3-pos MuSCs dictates muscle-specific regenerative capacity, with PAX3-pos cells driving rapid repair through enhanced activation, proliferation, and differentiation, while PAX3-neg cells replenish the quiescent reserve pool during repair. Notably, Pax3 is largely silenced in fetal progenitor cells that establish the adult muscle stem cell pool; however, a subset of these cells retains Pax3 expression into adulthood, raising important questions about their functional role (14). Similar to embryonic myogenesis, PAX3 and PAX7 display dynamic expression patterns during activation, echoing their cooperative yet stage-specific roles in MuSC regulation (33). The expression of Pax3 in activated myoblasts contrasts with its embryonic downregulation during the onset of differentiation, hinting at unique post-natal functions. Our findings show that Pax3 is required for maintaining a MuSC subset during muscle repair, as evidenced by the dramatic depletion of PAX3-pos MuSCs in injured BCPS muscles of Pax3cKO mice. This extends its known embryonic role in myogenic progenitor survival (10) to adult regeneration and reveals a context-dependent function, with defects confined to BCPS but not TA muscles. Moreover, the presence of abundant PAX3-pos MuSCs in the Rectus Femoris muscle supports that their spatial distribution is independent of embryonic origin and suggests that specific muscles may have been evolutionarily selected for PAX3-dependent regenerative advantage. Future work should address whether inducing Pax3 expression can enhance regenerative capacity in muscles with low endogenous PAX3 levels. The rapid cell death of PAX3-pos cells post-injury further underscores its role as a survival factor, preventing premature depletion of regenerative progenitors. In line with these observations, Morree et al. (35) demonstrated that MuSC-specific deletion of Pax3 in adult mice results in an increase in quiescence markers, a decrease in MYOD expression and EdU incorporation, and reduces myofiber fusion in diaphragm, gracilis, and triceps muscles under homeostatic conditions (35). Consistently, Pax3cKO mice displayed a decrease in grip strength and reduced performance in a treadmill assay, underscoring the role of PAX3 in the structural and functional homeostasis of the tissue (35). Additionally, Morree et al. (35) showed that Pax3 expression is post-transcriptionally regulated by alternative polyadenylation and by miR-206, which mediates MuSC activation in an injury-independent manner. Moreover, variable Pax3 expression levels underlie differing activation rates across muscle types (14). Consistently, our study uncovers a distinct transcriptional mechanism in which PAX3-dependent functional heterogeneity in MuSCs is defined not by transient marker fluctuations but by a fixed, genetically intrinsic stem cell population. Thus, our work extends beyond and differs fundamentally from previous studies, expanding the field’s understanding from regulation of quiescence to identity-based functional heterogeneity and regeneration rates. In our injury models, we observe a transient depletion of the PAX3-pos MuSC population during the early phases of regeneration followed by its restoration at later time points. These dynamics suggest that PAX3-pos and PAX3-neg MuSCs may contribute to tissue repair through temporally distinct cellular behaviors rather than through fixed lineage hierarchies. Lineage-tracing studies indicate that MuSC self-renewal occurs predominantly after approximately 5 dpi (36), whereas the early regenerative phase is largely devoted to progenitor expansion. Consistently, recent three-dimensional imaging analyses of regenerating muscle have shown that MuSC initially proliferate through planar divisions that expand the myogenic progenitor pool within the basal lamina niche (34). In this context, the transient reduction in the proportion of PAX3-pos MuSCs that we observe during early regeneration likely reflects their proliferative contribution to the expanding myoblast pool that fuel myoblast expansion and support efficient fiber regeneration, leading to their temporary dilution within the overall stem cell compartment. Within this framework, both PAX3-pos and PAX3-neg MuSC populations ultimately participate in rebuilding the stem cell reservoir, but they appear to contribute with distinct temporal division modes: PAX3-pos MuSCs preferentially support the early proliferative and regenerative phase, whereas other PAX3-neg MuSCs contribute more prominently to long-term maintenance of the stem cell compartment once tissue repair progresses. A key mechanistic insight from this work is the identification of SIX2 as a downstream effector of PAX3 in adult MuSCs. The exclusive expression of Six2 in PAX3-pos cells and its loss in Pax3cKO models position SIX2 as a critical mediator of PAX3 pro-regenerative effects. Functional assays confirmed that SIX2 restoration in PAX3-neg cells enhances proliferation, while its silencing in PAX3-pos cells disrupts differentiation. This mirrors the embryonic role of SIX2 in initiating the Myf5/Myod1 cascade (32, 37) and highlights its retained role in adult progenitors. Thus, the PAX3-SIX2 axis emerges as a conserved pathway between developmental and regenerative myogenesis. Building on the developmental described PAX3-SIX2 axis, we provide evidence that SIX2 fully rescues survival in Pax3-deficient MuSCs, restoring cell viability and preventing mitotic catastrophe (Fig. 5, C, E, and G). In addition, our RNA-seq analyses revealed that Six2 overexpression primarily activates cell cycle–related pathways and proliferation-associated genes (Fig. 4, K and L, fig. S5F), indicating that the primary transcriptional effect of SIX2 downstream of PAX3 is to promote MuSC proliferation. The enhancement in differentiation observed in our study, downstream of PAX3, is therefore likely an indirect outcome of increased progenitor expansion or is regulated by additional PAX3-dependent pathways influencing MuSC fate. Importantly, deeper analyses of the Six2 OE RNA-seq data in Pax3-deficient MuSCs reveals that this cell cycle reactivation is accompanied by concurrent upregulation of replication-coupled DNA repair pathways, including HDR through Homologous Recombination, Chromosome Maintenance, Resolution of D-loop Structures, and Regulation of TP53 Activity (fig. S5, G to K). The enrichment of these pathways in the Six2 OE suggests that Six2 actively rescues Pax3-deficient MuSCs from apoptosis toward checkpoint-competent cell cycle progression, as functinoally confirmed by the restoration of cell activation (EdU), viability (Tunel) and reduced DNA damage (γH2AX). Together, these results support a model in which PAX3 establishes a transcriptional program in this MuSC subpopulation that promotes survival and expansion during regeneration, at least in part through the activation of SIX2-dependent proliferative pathways. In our scRNA-seq dataset, the Pax3 transcript profile only partially overlaps with the Six2 expression pattern (Fig. 4, D and E). Because single-cell RNA sequencing measures transcript abundance rather than protein levels, these data may not fully capture the dynamics of PAX3 and SIX2 protein activity. One possible explanation is the presence of regulatory feedback within the PAX3-SIX2 axis. For instance, PAX3 may initially promote Six2 expression during MuSC activation, while subsequent regulatory mechanisms could modulate Pax3 levels as regeneration progresses. Regarding the molecular relationship between PAX3 and SIX2, while Six2 expression is lost in Pax3cKO MuSCs (Fig. 4F), demonstrating that PAX3 activity is required for normal Six2 expression, re-analysis of available chromatin occupancy datasets (38) did not identify direct PAX3 binding at the Six2 locus. This is consistent with an indirect regulatory network, in which PAX3 controls Six2 expression through transcriptional intermediaries rather than by direct promoter or enhancer binding. Such indirect transcriptional regulation is a well-established mechanism in developmental gene regulatory networks and does not diminish the functional significance of the PAX3-SIX2 axis. Indeed, work by Magli et al. (39) linked PAX3 chromatin occupancy to the regulation of Lbd1, supporting the existence of broader PAX3-dependent regulatory networks in myogenic cells that could mediate indirect control of Six2. The transcriptomic datasets provided in table S3 outline additional candidate pathways, including BMP and Notch signaling, that may mediate PAX3 regenerative effects and represent important directions for future works. Despite these insights, how Pax3 expression is dynamically regulated across muscles and whether its stress-protective mechanisms are conserved in aging or disease remain still open questions. Remarkably, Kirpatrick et al. (40) observed that PAX3-pos MuSCs decline in intrafusal fibers during aging, correlating with loss of fiber maintenance (40, 41). The decrease in MuSC numbers during aging was linked to weakened growth, regeneration, and repair (42, 43). Based on these studies, one can speculate that the progressive loss of PAX3-pos MuSCs drives this decline in MuSC number and function: if PAX3 were merely a developmental remnant, PAX3-pos and PAX3-neg MuSCs would decline at comparable rates. If, however, Pax3 expression is correlated with MuSC activation and proliferation, we could expect to see a greater decline in the proportion of PAX3-pos MuSCs as muscle ages. While our study uses Pax3 reporter mice to dissect MuSC heterogeneity, enabling lineage tracing analysis, technical limitations warrant careful interpretation. The Pax3CreERT2 (27) and Pax3nLacZ-IRES-GFP (16) are both heterozygous knock-in (KI) models, where exogenous cassettes are inserted into the endogenous Pax3 locus. This introduces the potential for haploinsufficiency and regulatory perturbations, as reporter expression relies on intact Pax3 regulatory elements and may not fully mirror the stability or dynamic expression of the endogenous protein, particularly under conditions of stress or regeneration. Thus, while these models are invaluable tools, these findings require validation in wild-type systems. However, the lack of PAX3-specific antibodies validated for in vivo use complicates MuSC characterization in wild-type models. As PAX3 and PAX7 share homologous DNA-binding domains and strongly conserved protein domains (38), commercial antibodies often exhibit cross-reactivity, necessitating reliance on transcriptional reporters like those employed in this study. While this study clarifies the role of PAX3 in the regeneration process, the basis for its muscle-specific activity remains unclear. Differences in embryonic origin (e.g., somitic vs. branchiomeric lineages) or local niche factors (e.g., BMP/Notch signalling gradients) could explain this disparity. Our observation leads to a conceptual questioning of the potential cross-talk between distinct MuSC subpopulations within specific muscle groups. Future work should map PAX3-pos MuSC distribution across muscle groups and define niche-derived signals regulating its expression and MuSC inter-communication. MATERIALS AND METHODS Mouse strains Mouse lines used in this study have been described and were provided by the corresponding laboratories: Pax3nLacZ-IRESGFP/+ (16), Pax3flox/+ (29), Rosa26flox(stop)flox-Tomato (R26stop-Tomato) (30), Pax3CreERT2/+ (27), Pax7CreERT2/+and R26RmTmG/+ (28). Animals were handled according to the European Community guidelines, implementing the 3Rs rule. Protocols were validated by the ethic committee of the French Ministry, under the reference numbers APAFIS #33241–202109261642469 v3 and APAFIS#94028379–2023122113266725-v3. Tamoxifen administration and muscle injury To induce recombination, mice from the Pax3creERT2;Pax3fl/fl;R26stop-Tomato and Pax3creERT2;Pax3fl/+;R26stop-Tomato control lines were fed with a solid tamoxifen diet (400 mg/kg of food) for 2 weeks and then kept 1 week with regular food. Mice were anesthetized with 3.5 μl/g of Isofluorane by inhalation. The biceps brachii, rectus femoris, tibialis anterior and gastrocnemius muscles were injected with 20 μl, 50 μl, and 70 μl (adjusted on muscle weight) respectively of 0.6% BaCl2 (Sigma, 202738). Western blot Muscles were dissected, minced and lysed in RIPA buffer (50 mM Tris HCl (Sigma, T5941), 150 mM NaCl (Sigma, 9625), 1% Igepal (Sigma, I8896), 0.5% Sodium deoxycholate (Sigma, D6750), 0.1% SDS (Sigma, L3771), 1 mM EDTA (Sigma, E5134), 1x Protease Inhibitor Cocktail (Roche, 04–693–116-001); pH8) for protein extraction. 20 μg of protein were used per sample. The blocking of the membrane was performed in 2% of gelatin from cold water fish (Sigma, G7765) and 0.2% Tween20 (Sigma, P9416) in PBS (Invitrogen, 003002) 1 h at room temperature. Primary antibodies were diluted in blocking solution and incubated 2 h at room temperature (table S1). Secondary antibodies were diluted in blocking solution and incubated 1 h at room temperature (table S1). Revelation reaction was performed 5 min at room temperature using SuperSignal West Pico (Thermo Fisher Scientific, 35060). Immunohistochemistry Cryosections of 10 μm were performed on biceps brachii, rectus femoris, tibialis anterior and gastrocnemius muscles previously dissected and frozen in liquid-nitrogen-cooled isopentane 2–4-7 or 28dpi. The muscle sections or the cultured cells underwent immunostaining as described (44) and as follows: fixed in 4% PFA for 15 min at room temperature, permeabilized in cold methanol for 6 min, boiled in citrate buffer (Dako, S1699) for epitope retrieval and blocked with 5% immunoglobin G (IgG)- free BSA (Jackson, 001–000-162) for 1 h at room temperature. To reduce background, the slides were incubated 30 min at room temperature with anti-mouse IgG Fab fragment (Jackson, 115–007-003). The sections were incubated with primary antibodies (table S1) diluted in 5% IgG-free BSA overnight at 4°C. Secondary antibody incubation (Invitrogen, Alexa-Fluor) was performed 1 h at room temperature followed by DAPI (Sigma, D9542) staining (1:5000) 10 min at room temperature (44). EdU staining reaction was performed according to manufacturer’s guidelines (Invitrogen, EdU Click-iT PLUS Kit, C10640). Similarly, Tunel staining reaction was performed according to manufacturer’s guidelines (Invitrogen, Click-iT PLUS kit TUNEL Assay, C10618). Muscle enzymatic dissociation Adult forelimb, trunk or hindlimb muscles were dissected, minced and incubated in digestion buffer (44) HBSS (Gibco, 14025092), 0.2% BSA (Sigma, A7906), 2 μg/ml Collagenase A (Roche, 11088793001), and 3.25 μg/ml Dispase II (Roche, 04942078001) for 2 h at 37°C. Following procedures previously described in (44),the muscle suspension was successively filtered through 100-μm cell strainers (BD Biosciences) and then spin at 600 g for 5 min at 4°C to remove large tissue fragments. The supernatant was collected, filtered through a 40 μm strainer and washed twice by centrifugation at 600 g for 5 min at 4°C. The final pellet was resuspended in cold HBSS (Gibco) supplemented with 0.2% bovine serum albumin (BSA, Sigma), and the cell suspension was used for fluorescence-activated cell sorting (FACS)-sorting. EdU in vivo assay For the EdU in vivo assay, fresh EdU (Sigma, T511285) solution at 6.67 mg/ml in 1 ml of PBS(1X) was prepared under the hood. The muscles of interest (BCPS and RF) were injured as previously described (see Tamoxifen administration and muscle injury section). 28 hours later, EdU solution was delivered by intraperitoneal injection (IP) in Pax3nLacZ-IRESGFP/+ (16) mice. The previously injured muscles were harvested 12-hours later. The isolated muscles were digested and stained with chicken polyclonal IgY anti-GFP antibody (see table S1). EdU labelling was performed according to the manufacturer’s guidelines (Thermo Fisher Scientific, EdU Click-iT kit for cytometry, C10635). MuSCs isolation by flow cytometry MuSCs were sorted from dissociated muscles with the cell sorter BD Influx Sorp (BD Biosciences) using a surface labelling with anti-mouse CD45-PE-Cy7, anti-Ter119-PE-Cy7, anti-mouse CD34-BV421, anti-mouse Sca1-FITC (all from BD Biosciences) and anti-mouse integrin-a7-A700 (R&D Systems) (45). Gating strategy for MuSCs isolation using cell surface markers was as follows: integrin-a7+, CD34+, Ter119-, CD45-, and Sca1-. In the mice Pax3nLacZ-IRESGFP/+, the PAX3-pos MuSCs were isolated based on GFP from the integrin-a7+, CD34+, Ter119-, CD45-, Sca1- antibodies and MuSC fraction. In the mice the Pax3creERT2;Pax3flox;RosaTdTomato PAX3-pos MuSCs were isolated based on TOM from the integrin-a7+, CD34+, Ter119-, CD45-, Sca1-, FITC- and MuSCs fraction. In the mice the Pax7CreERT2/+; RS26mTmG/+, PAX7-pos MuSCs were isolated based on TOM from the integrin-a7+, CD34+, Ter119-, CD45-, Sca1-, FITC- and MuSCs fraction. Isolated, mononuclear cells were collected in HBSS with0.2% BSA. Sorted cells were either plated or lysed for mRNA extraction following the protocols described below. Transplantation assay Immunodeficient Rag2−/−: γC−/− mice (8–12 weeks old) were used as transplantation hosts to prevent immune-mediated rejection of donor cells. Donor MuSCs were isolated from tamoxifen-induced Pax3CreERT2/+;R26stopTomato/+ mice (27) or not-induced Pax7CreERT2/+;R26RmTmG/+ mice, as previously described (25, 46). To create a regenerative niche conducive to MuSC engraftment, both TA muscles of recipient mice were injected with 50 μL of 0.6% BaCl2 (Sigma-Aldrich) under isoflurane anaesthesia, 24 hours prior to cell transplantation (19, 47). On the day of transplantation, MuSCs were isolated from donor mice using fluorescence-activated cell sorting (FACS). Forelimb and trunk muscles of Pax3CreERT2/+;R26stopTomato/+ mice (27) were enzymatically dissociated (collagenase A/dispase), and TOM-pos (tdTomato-expressing) cells were sorted as the PAX3-derived population. Hindlimb and trunk muscles from Pax7CreERT2/+;R26RmTmG/+ mice were processed similarly, and TOM-pos cells (from the mTmG reporter) were sorted as the PAX3-neg control population. After MuSCs isolation, for each recipient, 50,000 freshly sorted MuSCs were resuspended in 10 μl sterile FACS medium (HBSS+0.2%BSA) and injected into the right and left TA muscles using a Hamilton syringe. Right TA received PAX3-neg TOM-pos MuSCs, while left TA received PAX3-pos TOM-pos MuSCs. Six days post-transplantation, mice were euthanized, and TA muscles were harvested, frozen in liquid-nitrogen-cooled isopentane, and cryosectioned at 10 μm. Immunofluorescence staining was performed for donor-derived fibers and MuSCs, detected by TOMATO fluorescence. MuSC culture MuSCs isolated by flow cytometry were plated at 20,000 cells (proliferation assay), 40,000 cells (differentiation assay) per 1 cm2 and 1,000 cells (clonal expansion) per 10 cm2 on detachable 8 well-PCA treated m-slides (Sarstedt) pre-coated with Matrigel (Corning, 354248) for 30 min at 37°C and cultured in DMEM (Gibco, 41966029) with 20% FCS (Eurobio, CVFSVF00–01), 1% Pen/Step (Gibco, 15070063), and 4 ng/ml basic FGF (bFGF) (Peprotech, 450–33). In vitro gain- and loss-of-function experiments PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs were isolated by FACS and cultured in growth medium. For SIX2 knockdown, MuSCs were transfected with 50 nM Six2 siRNA (Horizon Discovery, catalog #L-042261-01-0005) or non-targeting (control) siRNA (Horizon Discovery, negative control, catalog #D-001810-01-05). Lipofectamine 3000 (Thermo Fisher Scientific, L3000008) was used according to the manufacturer’s protocol. Cells were incubated at 37°C, 5% CO2 for 72 hours before downstream assays (RT-qPCR, proliferation, or differentiation analysis). Transfection efficiency was monitored by co-transfection with a TOMATO reporter and knockdown efficiency was confirmed by RT-qPCR for Six2 mRNA. Functional assays (proliferation, differentiation) were performed 48 and 72 hours post-transfection. To overexpress Six2, PAX3-pos (GFP-pos) and PAX3-neg (GFP-neg) MuSCs were transfected with 2 μg of pCR3-Six2 plasmid containing the mouse Six2 cDNA (48), or control, empty pCR3 plasmid (Clontech). Lipofectamine LTX with PLUS reagent (Thermo Fisher Scientific, 15338100) was used according to the manufacturer’s protocol. Transfection efficiency was verified by RT-qPCR for Six2 mRNA. Induction of Pax3 deletion in vitro 24 hours after plating, FACS-isolated MuSCs from Pax3creERT2;Pax3fl/fl;R26stop-Tomato and Pax3creERT2;Pax3fl/+;R26stop-Tomato control mice were treated with 2 μM 4-hydroxytamoxifen (4-OHT) (Sigma, H7904) for 48 hours to induce CRE-mediated recombination and Pax3 excision, following protocols adapted from previous studies (44). Recombination efficiency was assessed by RT-qPCR for Pax3 and by TOMATO fluorescence. SIX2 gain-of-function rescue Following confirmation of Pax3 downregulation, MuSCs were transfected with the pCR3-Six2 (48) plasmid or an empty vector control using Lipofectamine LTX with PLUS reagent (Thermo Fisher Scientific, 15338100) according to the manufacturer’s instructions. Transfection efficiency was monitored by co-transfection with a GFP reporter and confirmed by RT-qPCR for Six2 expression levels. RNA extraction, cDNA synthesis and RT-qPCR A minimum of 50,000 cells, cultured in high serum-containing medium and at low density, were collected per sample for RNA extraction (Quick-RNA, Zymogen, ZR1050) following the manufacturer’s protocol. Reverse transcription was performed using the SuperScript III Reverse Transcriptase (Invitrogen, 18080–093) following the manufacturer’s guidelines. RT-qPCR was performed using the PowerUp SYBR Green Master Mix (Applied Biosystems, A25742). The relative mRNA levels were calculated using the 2^−ΔΔCt method. The ΔCt were obtained from Ct normalized to the housekeeping gene Tbp levels in each sample. The RT-qPCR primers used are listed in table S2. RNA-sequencing RNA concentrations were assessed using the Qubit RNA HS Assay Kit (Thermo Fisher Scientific, Q32855) on a Qubit 3.0 Fluorometer (Life Technologies). cDNA was generated following the SMART-Seq2 protocol (49). In brief, Superscript II Reverse Transcriptase was used for reverse transcription, and 12 cycles of PCR pre-amplification were performed with the KAPA HiFi HotStart ReadyMix (Kapa Biosystems, KK2602). cDNA was purified by AMPure XP beads (Beckman Coulter), and the concentration was assessed using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Q32854). Tagmentation of the amplified cDNA was performed with Tn5 transposome using the TruePrep DNA Library Prep Kit V2 for Illumina (Vazyme, TD501), followed by 12 cycles of library amplification with the KAPA HiFi PCR Kit (Kapa Biosystems, KK2102). Size selection was carried out using SPRIselect reagent (Beckman Coulter): a 0.55X bead ratio was first applied to remove fragments greater than 500 bp, and the supernatant was subsequently subjected to a 0.7X bead ratio to retain fragments in the 200–500 bp range. Tagmented DNA concentration and average fragment size were verified using the Qubit dsDNA HS Assay Kit and DNF-474 HS NGS Fragment Kit (Agilent) on the Fragment Analyzer System (Agilent), respectively. Libraries were pooled and converted to circular ssDNA using the MGIEasy Universal Library Conversion Kit (App-A) (MGI, 1000004155) following the manufacturer’s instructions. Sequencing was performed on the DNBSEQ-T7 platform (paired-end, 150 bp). The Fastq files were mapped to the mouse reference genome GRCm38 using STAR (v2.7.11a). Quantification was conducted with RSEM (v1.3.3) based on the Gencode vM25 annotation. Quality control was carried out using FastQC (v0.12.1) and aggregated with multiQC (v1.19). Raw gene counts were normalized and analyzed for differential expression using DESeq2 (v1.42.0). Pathway enrichment was analyzed using GSEA (Gene Set Enrichment Analysis) from the clusterProfiler package (v4.18.4) in R (50). Single-cell RNA sequencing (scRNA-seq) experiment and analysis Freshly isolated PAX3-pos(GFP-pos) and PAX3-neg(GFP-neg) MuSCs from biceps brachii and rectus femoris muscles 5-days-post-injury, were processed for scRNA-seq and 20,000 cells per condition were loaded into the 10× Chromium Chip with the Single Cell 3′ Reagent Kit v3.1 (reference 1000268) according to the manufacturer’s protocol. Four biological samples of each condition were pooled before making the libraries. Libraries were sequenced by pair with a HighOutput flowcell using an Illumina Nextseq 500. A minimum of 50,000 reads per cell were sequenced and analysed with Cell Ranger Single Cell Software Suite by 10× Genomics. Raw base call files from the Nextseq 500 were demultiplexed with the cellranger mkfastq pipeline into library-specific FASTQ files. The FASTQ files for each library were processed independently with the cellranger count pipeline. This pipeline used STAR to align cDNA reads to the Mus musculus genome reference (mm10)-2020-A. The Seurat package (v3.0) (51) under R (v3.6.1) (R Core Team (2024). R: A language and environment for statistical computing. R Foundation for Statistical Computing https://www.gbif.org/fr/tool/81287/r-a-language-and-environment-for-statistical-computing) was used to perform downstream clustering analysis on scRNAseq data as previously described (52). Cells went through a classical Quality Control using the number of detected genes per cell (nFeatures), the number of mRNA molecules per cell (nCounts) and the percentage of expression of mitochondrial genes (pMito) as cut-offs. Outliers on a nFeature vs nCount plot were manually identified and removed from the dataset. Gene counts for cells that passed the above selections were normalised to the total expression and Log-transformed with the NormalizeData function of Seurat using the nCount median as a scale factor. Highly variable genes were detected with the FindVariableFeatures function (default parameters). Using highly variable genes as input, principal component analysis was performed on the scaled data to reduce dimensionality. Statistically significant principal components were determined by using the ElbowPlot function. Cell clusters were generated with the FindNeighbors/FindClusters functions (default parameters except for the number of selected PCs). Different clustering results were generated at different resolutions and for different sets of PCs. Non-linear dimensional reduction (UMAP) and clustering trees using Clustree were used to visualise clustering results and select the most robust and relevant result. Markers for each cluster were found using the FindAllMarkers function of Seurat (using highly variable genes as an input, default parameters otherwise) that ran Wilcoxon rank-sum tests (adjusted P value < 0.05). Image capturing, quantification, and statistical analysis Image capturing was performed using a Zeiss LSM 800 confocal microscope with the associated Zeiss Zen Lite v2.3software. Quantifications were performed in at least six pictures taken randomly within each sample, per experiment, using Image J and statistics were done with GraphPad Prism version 10 and Excel. The statistical test performed in each analysis is described in the associated figure legend.