Restoration of functional PAX3 transcriptional factor enhanced neuronal differentiation in PAX3b isoform-depleted neuroblastoma cells

Reexpressed PAX3 transcription factor is believed to be responsible for the differentiation defects observed in neuroblastoma. Although the importance of PAX3 in neuronal differentiation is documented how it is involved in the defective differentiation remains unexplored particularly with its isoforms. Here, first we have analyzed PAX3 expression, its functional status, and its correlation with the neuronal marker expression in SH-SY5Y and its parental SK-N-SH cells. We have found that SH-SY5Y cells which expressed more PAX3 showed increased expression of neuronal marker genes (TUBB, MAP2, NEFL, NEUROG2, SYP) and reported PAX3 target genes (MET, TGFA, and NCAM1) than the SK-N-SH cells that had low PAX3 level. Retinoic acid treatment is unable to induce neuronal differentiation in cells (SK-N-SH) with low PAX3 level/activity. Moreover, ectopic expression of PAX3 in SK-N-SH cells neither induces neuronal marker genes nor its target genes. PAX3 isoform expression analysis revealed the expression of PAX3b isoform that contains only paired domain in SK-N-SH cells, whereas in SH-SY5Y cells, we could also observe PAX3c isoform that contains all functional domains. Further, PAX3b depletion in SK-N-SH cells is not induced PAX3 target genes, and the cells remain poorly differentiated. Interestingly, ectopic PAX3 expression in PAX3b-depleted SK-N-SH cells enhanced neuronal outgrowth along with neuronal marker gene induction. Collectively, these results showed that the PAX3b isoform may be responsible for the differentiation defect observed in SK-N-SH cells and restoration of functional PAX3 in the absence of PAX3b can induce neurogenesis in these cells.


Introduction
Neuroblastoma is the most common extra-cranial childhood tumor that arises from the sympathoadrenal lineage of neural-crest cells and is often diagnosed in the first year of life (Johnsen et al. 2019). The primary neuroblastomas usually occur in the adrenal medulla, and it is also located in the neck, chest, or pelvis (Papaioannou and McHugh 2005). The clinical outcome of neuroblastoma is highly variable with inconsistent morbidity and mortality (Louis and Shohet 2015). Complete chromosome gains are associated with a favorable neuroblastoma prognosis and segmental chromosomal aberrations including MYCN amplification, regional chromosomal gains, or losses associated with its poor outcome (Pinto et al. 2016). Recent advancements in neuroblastoma diagnostics and treatment strategies improved the overall survival of the lower-risk groups; however, for the high-risk group, it remains poor (Tolbert and Matthay 2018). Conventional chemotherapeutic drugs are highly toxic and nonspecific and provide severe side effects in children which necessitates the requirement of safe therapeutics (Ruggiero et al. 2018).
Pediatric tumors including neuroblastoma hypothesized to originate from defective cellular differentiation due to their embryonal nature (Ponzoni et al. 2022). Induction of differentiation in tumor cells is considered to be a safe therapeutic strategy, and it requires a thorough understanding of differentiation defects at a molecular level (Enane et al. 2018). Overexpression of developmentally important transcription factors is frequently reported in childhood malignancies, and they are believed to be involved in defective differentiation observed in tumor cells (Vishnoi et al. 2020;Islam et al. 2021). PAX3 belongs to paired box family of transcription factors involves in the various cellular process including cellular proliferation, migration, lineage specificity, and differentiation (Boudjadi et al. 2018). During embryonic development, PAX3 contributes to diverse cellular lineages including neural crest-derived peripheral nervous system, and its downregulation is essential for neuronal progenitor cells to undergo differentiation (Monsoro-Burq 2015). Several studies from mice and chick embryo showed a positive role of PAX3 in neurogenesis (Mansouri and Gruss 1998;Agoston et al. 2012;Adams et al. 2014;Lin et al. 2016). In contrary, the negative role of PAX3 in neurogenesis was also reported in neural stem cells where PAX3 suppressed TUBB3 expression (Cao et al. 2017). Further PAX3 expression and acetylation promoted neural stem cell proliferation (Nakazaki et al. 2008;Ichi et al. 2011). Reexpression of PAX3 is frequently reported in neuroblastoma, rhabdomyosarcoma, Ewing's sarcoma, Wilm's tumor, and glioma or its derived cell lines (Schulte et al. 1997;Hueber et al. 2011;Zhu et al. 2018). A functional study using siRNA-mediated PAX3 inhibition showed reduced growth, migration, and invasion in neuroblastoma cell lines and also showed enhanced cytotoxicity to chemotherapeutic drugs (Fang et al. 2014). However, a recent PAX3 overexpression study showed inhibition of cell proliferation and viability in neuroblastoma cell line Neuro-2A (Huo et al. 2021). The above-mentioned studies show the contrary roles of PAX3 in neuronal differentiation and neuroblastoma development, and the exact contribution remains unexplored.
There are eight PAX3 isoforms (PAX3a, PAX3b, PAX3c, PAX3d, PAX3e, PAX3g, PAX3h, and PAX3i) are reported (Boudjadi et al. 2018), and they showed a different effects on proliferation, apoptosis, and migration in melanocytes (Wang et al. 2006). Among these isoforms, PAX3a and PAX3b have only paired and no homeo and transactivation domains (Tsukamoto et al. 1994). PAX3c (a well-studied isoform of PAX3), PAX3d, and PAX3e have all three functional domains, and PAX3g, PAX3h, and PAX3i have truncated transactivation domains (Tsukamoto et al. 1994;Boudjadi et al. 2018). PAX3a and PAX3b expression has been reported in the human brain and cerebellum; however, their role in establishing these tissues is not documented (Tsukamoto et al. 1994;Thompson et al. 2021). Hence, the tissue-specific expression of various PAX3 isoforms and their functional role is limited in the literature. Particularly, the role of PAX3 isoforms in neuronal differentiation and its defects in neuroblastoma remains unexplored. Therefore, in this current study, we have focused on a detailed understanding of PAX3 and its isoform's role in neuroblastoma cell differentiation. To explore the differentiation defect in neuroblastoma cells, we have used SK-N-SH and its derived SH-SY5Y cell lines. Although both cell lines are derived from the same patient, SH-SY5Y cells show more neuronal phenotypes than the parental SK-N-SH cells and are widely used for neuronal differentiation analysis (Kovalevich and Langford 2013). Here we have analyzed the PAX3 expression and activity and correlated it with neuronal marker genes expression and retinoic acid-mediated differentiation induction in SK-N-SH and SH-SY5Y cells. Further, we have found that ectopic PAX3 expression or PAX3b isoform depletion is unable to rescue the neuronal differentiation defect in SK-N-SH cells. However, ectopic PAX3 expression induced a neuronal phenotype in PAX3b-depleted SK-N-SH cells. Hence, our data demonstrated that the PAX3b isoform may obstruct PAX3-mediated neuronal differentiation in neuroblastoma cells, and restoration of functional PAX3 can induce neuroblastoma cell differentiation in the absence of PAX3b.

Cell lines
The neuroblastoma cell lines SH-SY5Y, SK-N-SH, and 293T cells were acquired from the National Center for Cell Science, Pune, India. All the cells were cultured in DMEM medium with 10% fetal bovine serum and antibiotic-antimycotic solution (Gibco) at 37 °C, 5% CO2 in a humidified atmosphere. For the retinoic acid treatment, the cells were incubated with 10 µM retinoic acid in serum deprival condition as differentiation media (DM) for 4 days.

Plasmids
shRNA targeting PAX3b and scramble control were used for depletion experiments in neuroblastoma cells. A sense and antisense oligonucleotides mentioned in the Supplementary Table 1 were used to construct PAX3b and scramble shRNAs. The sense and antisense oligos were annealed by incubating at 95 °C for 5 min followed by slow cooling to room temperature. The annealed oligos were cloned into the pLKO.1-TRC1 plasmid vector, a gift from David Root (Addgene plasmid #10,878) into EcoRI and AgeI sites; pBabe-PAX3 and viral packaging plasmids (kind gift from Dr. Asoke Mal) were used in this study for the transient or stable expression of PAX3 (Jothi et al. 2012).

Western blot analysis
Protein expressions were analyzed in western blot analysis. For this the total protein extracts were obtained by 1 3 resuspending the cells cultured under specific conditions in lysis buffer (50 mM Hepes, 150 mM sodium chloride, 0.1% NP-40, 1 mM EDTA, 5 mM tetra sodium pyrophosphate, 10% glycerol, 25 mM β-glycerophosphate, 20 mM sodium fluoride, 0.2 mM sodium orthovanadate, and 1 × Protease inhibitor). After 20-min incubation in ice, the cells were sonicated for 30 s at 50% amplitude in Qsonica Q125 sonicator. Then the crude lysates were centrifuged at 12,000 rpm for 10 min at 4 °C. The protein present in the total cellular extracts was measured by a Protein assay reagent (Bio-Rad). Forty to eighty microgram of total cell extracts were separated in a SDS-PAGE gel and transferred to PVDF membrane by wet transfer methods. After transfer, the membranes were blocked with blocking buffer containing 5% nonfat dry milk for 1 h and washed twice in TBST and once in TBS. The primary antibodies were diluted according to manufacturer recommendation and incubated the membrane for overnight. The next day, the membranes were washed in TBST for twice and once in TBS and then incubated with HRP-conjugated secondary antibody for 1 h in room temperature. Then the bound antibodies were detected using HRP substrate (Luminata Forte Westerns HRP substrate, Merck Millipore). The resulting blots were visualized by the ChemiDoc XRS + imaging system (Bio-Rad). Antibody information is given in Supplementary Table 2.

RNA expression analysis
Total RNAs from cells were isolated by the TRIzol™ (Invitrogen) method followed by the cDNA synthesis using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific™) according to the manufacturer's instruction. The resulted cDNA was diluted to a 1:5 ratio with nucleasefree water and used for quantitative/semiquantitative RT-PCR. RT-qPCR was performed using the Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific™) in the QuantStudio™ 6 Flex Real-Time PCR System (Applied Biosystems™) using the initial denaturation at 95 °C for 5 min followed by 40 cycles of denaturation at 95 °C for 1 min, annealing at 55 °C for 30 s, and extension of 72 °C for 1 min with a final extension step at 72 °C for 2 min. Gene expression quantification of triplicate experiments were performed using the delta-delta CT method, and the qPCR amplicon specificity was verified by melting curve analysis. For semiquantitative PCR, PCR master mix (Thermo Sci-entific™) was used to amplify cDNA products under the above-mentioned condition. The amplicon was loaded onto agarose gel, and images were acquired from the ChemiDoc XRS + imaging system, Bio-Rad. Quantification of PCR amplicons was performed by the ImageJ software version 1.53q. Primer details are given in Supplementary Table 1.

Immunofluorescence
For immunofluorescence, the cells grown in 35-mm culture dishes were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton X-100 in PBS for 30 min. After blocking with 3% horse serum, the cells were washed with PBS containing 0.1% Triton X-100 and incubated overnight with indicated antibodies. The next day, the unbound antibodies were washed two times with 0.1% Triton X-100 in PBS. Then the bound primary antibodies were detected using an Alexa Fluor-conjugated secondary antibody for 1 h incubation. After washing the cells, the nuclei were counterstained with DAPI, and the images were acquired with a fluorescence microscope (Olympus IX73). Antibody information is given in Supplementary Table 2.

Viral preparation and transduction
The viral protocol was approved by the Institutional Biosafety Committee, NIMHANS (NIMHANS/DO/IBSC MEETING/2019/3). The day before transfection, 293T cells were cultured in a 35-mm plate to reach a confluency of 80% at the time of transfection. The cells were transiently transfected with pLKO.1-TRC-ShPAX3b or pBabe-PAX3 or control plasmids along with respective viral packaging vectors using the calcium phosphate coprecipitation method. We have used 12.5 mM calcium chloride and 2XHBS solution to transfect 8 µg of total plasmids for a 35-mm plate. Two days after transfection, the virus-containing supernatants were collected and filtered with 0.45-μm syringe filters. For the transduction, these viruses were diluted in a growth medium with 8 µg/ ml of polybrene. After 2 days, virus-transduced cells were selected with 3 µg/ml of puromycin for at least 5 days and pooled.

Neuronal outgrowth measurements
Cells containing neurites were traced from microscopic images of each experimental group using the ImageJ software (NIH, MD, USA). Neurite with double the volume of the cell body was considered for analysis. The total neurite length and the number of neurites in each group were analyzed.

Statistics
Results were expressed as mean ± standard error (mean ± s.e). Statistical analysis was performed by Student's t-test. A level of p < 0.05 was considered to be significant.

Poor neuronal features in neuroblastoma cells with low PAX3 expression
To explore the differentiation defect in neuroblastoma cells, we have used SK-N-SH and its derived SH-SY5Y cell lines (Kovalevich and Langford 2013). SH-SY5Y cells reported to show more neuronal phenotypes than the parental SK-N-SH cells (Kovalevich and Langford 2013). To further confirm this, we have analyzed the neuronal features of these cells by immunostaining and blotting with neuronal marker gene class III beta 3 tubulin (TUBB3) which is involved in axon guidance during neurogenesis. Immunofluorescence results showed more neuron-like features in SH-SY5Y cells than in SK-N-SH (Fig. 1a, b). The western blot analysis revealed the increased expression of TUBB3 in SH-SY5Y cells than in SK-N-SH (Fig. 1c, c'). RT-qPCR analysis of neuronal marker genes including TUBB3, MAP2 (microtubule-associated protein 2) involved in microtubule assembly, NEFL (neurofilament light chain) that maintains neuronal caliber, NEUROG2 (neurogenin 2) that specifies neuronal fate, and SYP (synaptophysin) a part of synaptic vesicles showed increased expression in SH-SY5Y cells than SK-N-SH cells (Fig. 1d). Thus, these data confirm that SH-SY5Y cells were more neurogenic than the parental SK-N-SH cells that showed poor neuron-like features ( Fig. 1a-d).
We have analyzed PAX3 expression at RNA and protein levels and its functional status in both cells. Here we have found an increased level of PAX3 mRNA and protein in SH-SY5Y compared to SK-N-SH cells (Fig. 1e, f, f'). PAX3 functional status is assessed by the expression of its reported target genes (TGFA, MET, and NCAM1). The results revealed the expression of PAX3 target genes in SH-SY5Y, and their level was relatively lower in SK-N-SH (Fig. 1g). Thus, we observed poor neuron-like phenotypes in SK-N-SH with low PAX3 levels compared to SH-SY5Y with functional PAX3 (Fig. 1).

Retinoic acid induces neuronal differentiation in high PAX3 expressing neuroblastoma cells
Retinoic acid (RA) is a well-known agent to induce neuronal differentiation in various cells (Xun et al. 2012;Tan et al. 2015;Park and Rhee 2018). We have analyzed the effect of retinoic acid in inducing neuronal differentiation in these neuroblastoma cell lines. Firstly, we have used immunofluorescence to detect TUBB3 expression and assess the morphological changes by RA treatment (Fig. 2a-f). In SH-SY5Y cells, TUBB3 expression was found in both GM and DM conditions, whereas in SK-N-SH cells, we observed its reduced level in DM conditions, and it is also reflected in western blot analysis  Fig. 2a-d'', g, g'). The results also revealed that RA can induce neuronal outgrowth in SH-SY5Y not in SK-N-SH cells under DM conditions (Fig. 2a-d''). Quantification of neurite length and the number of cells with two or more neurites confirmed RA-mediated induction of neuron-like morphological features only in SH-SY5Y cells (Fig. 2e, f). Then mRNA expression analysis showed induction of TUBB3, MAP2, and NEFL and reduction of SYP in RA-treated SH-SY5Y cells and no difference in NEUROG2 expression (Fig. 2h). On the contrary, none of these neuronal marker genes were induced in SK-N-SH cells by RA (Fig. 2i). Collectively, these evidences suggest that RA can enhance neurogenic differentiation in PAX3 functional SH-SY5Y cells not in SK-N-SH that had low PAX3 level/activity (Figs. 1 and 2).

Ectopic PAX3 expression fails to induce neurogenesis in SK-N-SH cells
Here we tested whether ectopic expression of PAX3 can enhance neuronal differentiation in SK-N-SH cells or not.

Fig. 2 Retinoic acid unable to induce neuronal differentiation in SK-N-SH cells.
Immunofluorescence showing TUBB3 expression in neuroblastoma cells treated with (GM) or without (DM) retinoic acid (a-d). Neuronal outgrowth by RA was quantified as mean neurite length (e) and the number of cells with two or more neurites (f). TUBB3 protein expression and its quantification were shown in g and g'. Neuronal marker genes expression in SH-SY5Y (h) and SK-N-SH (i) under GM and DM conditions. Data are shown as mean value ± s.e.m, n = 3. nd, not detectable; p value * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001; ns, nonsignificant. Scale bar 50 μM For this, we have generated stable SK-N-SH cells that express PAX3 cDNA (SK-PAX3) or vector (SK-vec). After confirmation of ectopic PAX3 in these cells (Fig. 3a, a'), we have analyzed the neurogenic features by TUBB3 immunostaining, and the results revealed slightly improved neuronal outgrowth (Fig. 3b-e). However, western blot analysis showed no changes in the expression of TUBB3 in both SK-PAX3 and SK-vec cells (Fig. 3f, f'). Further, the expression of other neuronal marker genes also showed no notable change in these cells (Fig. 3g). Moreover, the ectopically expressed PAX3 in these cells was functionally inactive as shown by its target gene expression (Fig. 3h). Thus, these results revealed that ectopic PAX3 in SK-N-SH is unable to rescue neuronal differentiation due to its functional inactivation (Fig. 3).

Depletion of PAX3b isoform alone is not sufficient to induce neuronal differentiation in SK-N-SH cells
There were eight PAX3 protein isoforms identified, and they varied in their structural features (Boudjadi et al. 2018). The inactivation status of PAX3 observed in SK-PAX3 cells (Fig. 3h) could be due to interference of transcriptionally inactive PAX3 isoforms. To verify this first, we have analyzed the expression of various PAX3 isoforms in SH-SY5Y and SK-N-SH cells using semiquantitative RT-PCR. Here, in SH-SH5Y cells, we could observe both PAX3b and PAX3c isoform expression, whereas in SK-N-SH cells, we could see only PAX3b expression (Fig. 4a, a'). We could not detect any other PAX3 isoforms in these cells (Fig. 4a, a').
Defective neuronal features observed in SK-N-SH cells (Fig. 1) could be due to the presence of PAX3b (Fig. 4a, a'). To verify this, we generated stable PAX3b-depleted SK-N-SH (SK-shPAX3b) and scramble control (SK-shScr) cells by lentiviral-mediated transduction, and the PAX3b depletion was confirmed by semiquantitative RT-PCR analysis (Fig. 4b, b'). Immunofluorescence analysis showed slightly enhanced neuronal morphology in SK-shPAX3b compared to SK-shScr cells (Fig. 4c-f) with no change in TUBB3 expression, and the same is confirmed in western blot analysis (Fig. 4g, g'). RT-qPCR analysis showed a reduction of all of the neuronal markers in SK-shPAX3b compared to control cells except NEUROG2 (Fig. 4h). PAX3 target genes were also reduced in PAX3b-depleted cells compared to scramble control (Fig. 4i). These data support that PAX3b-depletion  (a, a'). TUBB3 expression (b-c) and neuronal outgrowth quantification (d and e) in SK-vec cells and SK-PAX3 were shown. TUBB3 protein expression in these cells is shown in f and f'. RT-qPCR showing neuronal markers (g) and PAX3 targets (h). Data are shown as mean value ± s.e.m, n = 3. nd, not detectable; p value * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001; ns, nonsignificant. Scale bar 50 μM alone is not sufficient to induce neuronal differentiation in SK-N-SH cells (Fig. 4).

Functional PAX3 can induce neuronal differentiation in PAX3b-depleted SK-N-SH cells
The defect in inducing neuronal differentiation in SK-shPAX3b cells could be due to the absence of functional PAX3 (Fig. 4), and the restoration of functional PAX3 can induce neuronal differentiation in these cells. To test this, we have ectopically expressed PAX3 cDNA or vector into SK-shPAX3b cells. The ectopic PAX3 expression in SK-shPAX3b cells was confirmed by western blot analysis (Fig. 5a, a'). Ectopic PAX3 induced expression of PAX3 target genes in SK-shPAX3b cells (Fig. 5b). Immunofluorescence analysis confirmed the increased expression of TUBB3 and morphological differentiation in PAX3-transfected SK-shPAX3b cells (Fig. 5c-d''). We have also noticed increased neuronal outgrowth in PAX3 ectopically expressed PAX3b-depleted cells (Fig. 5e, f). Western blot analysis also confirms the induced expression of neuronal marker TUBB3 in PAX3-transfected cells (Fig. 5g, g'). Besides, we have observed the induction of neuronal marker genes in these cells (Fig. 5h). Overall, these data suggest that restoration of functional PAX3 in PAX3b-depleted SK-N-SH cells can induce neuronal differentiation (Fig. 5).

Functional PAX3 is associated with neuroblastoma cells differentiation
In this study, we report that the establishment of functional PAX3 in the neuroblastoma cells can induce neuronal differentiation in the absence of PAX3b isoform. The importance of PAX3 in neuronal differentiation is explained in several studies (Dude et al. 2009;Ichi et al. 2011;Agoston et al. 2012;Blake and Ziman 2013;Adams et al. 2014). Further, PAX3 is implicated in neural tube development, Fig. 4 PAX3b depletion is not sufficient to restore neuronal differentiation in SK-N-SH cells. mRNA expression of various PAX3 isoforms in SH-SY5Y and SK-N-SH cells were shown in a and a'. PAX3b isoform depletion in SK-shPAX3b is confirmed in the RT-PCR (b and b'). Immunofluorescence showing TUBB3 expression (c and d) and neurites quantification (e and f) in PAX3bdepleted cells. TUBB3 protein expression (g and g'), mRNA expression of neuronal marker genes (h), and PAX3 target genes (i) in PAX3b-depleted SK-N-SH cells. Data are shown as mean value ± s.e.m, n = 3. nd, not detectable; p value * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001; ns, nonsignificant. Scale bar 50 μM neuronal differentiation of sensory, and commissural neurons of chick and mice spinal cord development as has been reported earlier (Mansouri and Gruss 1998;Lin et al. 2016). Contrarily, a set of studies showed the negative implication of PAX3 in neuronal differentiation (Cao et al. 2017;Huo et al. 2021). Increased expression of PAX3 were reported in many tumors including neuroblastoma which often shows defective differentiation; however, its functional status and correlation with the differentiation potential are not available (Fang et al. 2014). Here among SK-N-SH and SH-SY5Y neuroblastoma cells, we have found that SH-SY5Y shows more neurogenic features than the parental cells (Fig. 1a-d) (Kovalevich and Langford 2013). The improved neuronal features observed in SH-SY5Y cells could be due to increased PAX3 level/activity, whereas its low level might contribute for the poor differentiation phenotype found in SK-N-SH cells (Fig. 1). Retinoic acid (RA) is a well-known agent used to induce neuronal differentiation in various cells including neuroblastoma cells (Sidell et al. 1983;Xun et al. 2012;Tan et al. 2015;Park and Rhee 2018). Due to the neurogenic properties of RA, it is included in the current treatment plan for neuroblastoma patients (Matthay et al. 2009). However, RA treatment is not beneficial to high-risk neuroblastoma patients (Reynolds et al. 2003). This inconsistent response by RA treatment is indicative of its variable effect on tumor cells. Here we have noticed RA-mediated induction of neuronal differentiation in SH-SY5Y cells, and we did not observe any such neurogenic conversion in SK-N-SH cells which showed low PAX3 level/activity (Fig. 2). Similarly, the earlier research work also showed no induction of neuronal phenotype upon RA treatment in SK-N-SH cells; rather it converts these cells into epitheliallike morphology (Sidell et al. 1983). In contrast to these observations, the recent article claimed that RA treatment induced neuronal differentiation in SK-N-SH cells as shown by neuronal marker SYP expression (Illendula et al. 2020).  a and a'). PAX3 target genes expression in ectopically PAX3 expressed SK-shPAX3b cells shown in b. TUBB3 expression and neuronal outgrowth were shown in immunofluorescence image (c and d) and its quantification (e and f). PAX3 ectopic expression induced TUBB3 protein (g and g') and neuronal marker genes expression (h) in SK-shPAX3b cells. The possible role of PAX3 and PAX3b in neuroblastoma cell differentiation is shown as schematic image (i). Data are shown as mean value ± s.e.m, n = 3. nd, not detectable; p value * ≤ 0.05; ** ≤ 0.01; *** ≤ 0.001; **** ≤ 0.0001; ns, nonsignificant. Scale bar 50 μM However, their data showed more than 50% reduction of SYP in 48 h of differentiation which is in contrast to their claim (Illendula et al. 2020), and we have reported a similar reduction of TUBB3 by RA treatment in SK-N-SH cells (Fig. 2g, g', i). Further, another study by Liang et al. (2009) also could not find the RA-mediated induction of TUBB3 or SYP in SK-N-SH cells. Nevertheless, our observations support that RA can induce neuronal differentiation in cells with functional PAX3 (Fig. 2). These evidences along with existing data support that PAX3 function is necessary for the neuronal differentiation, and its function may be compromised during neuroblastoma development which is responsible for the differentiation defect observed in neroblastoma cells.

Restoration of functional PAX3 induced neuroblastoma cells differentiation in the absence of PAX3b isoform
Several evidences showed the importance of PAX3 in neuronal differentiation (Dude et al. 2009;Ichi et al. 2011;Agoston et al. 2012;Blake and Ziman 2013;Adams et al. 2014). However, ectopic PAX3 expression failed to induce expression of neuronal markers although it showed slight morphological differentiation in SK-N-SH cells and these unpredicted results may be due to its functional inactivation as shown by its target gene expression (Fig. 3). Eight PAX3 isoforms with variable transactivation potentials were identified in humans, and they have shown different effects on tumorigenic properties in melanoma cells (Wang et al. 2006;Boudjadi et al. 2018). There is no detailed study available on PAX3 isoforms in neuronal differentiation in human cells, and mouse does not have all these isoforms. There is one study that revealed the role of different PAX3 isoform in chick sensory neurogenesis (Adams et al. 2014). In this study, we have found that PAX3b isoform contains only paired DNA-binding domain in both SK-N-SH and SH-SY5Y cells, and the well-characterized PAX3c isoform contains all functional domains in SH-SY5Y cells (Fig. 4a,  a'). PAX3 target gene expression is lower in SK-N-SH than in SH-SY5Y cells, and the ectopic PAX3 is also unable to induce these target genes in SK-N-SH cells (Figs. 1g and  4a). Interference of endogenous PAX3b isoform with ectopic PAX3 function in SK-N-SH cells might be responsible for the failure to induce neuronal differentiation ( Fig. 3b-g). Further, PAX3b depletion alone is not sufficient to induce neuronal differentiation in these cells, and it requires restoration of functional PAX3 (Fig. 4). Interestingly, ectopic PAX3 expression in PAX3b-depleted SK-N-SH cells completely restored the PAX3 activity (Fig. 5a, b), and the detailed mechanism needs to be explored further. As expected, the restoration of functional PAX3 enhanced neuronal features like an increased number of dendritic projection and their length along with induction of neuronal marker genes in PAX3b-depleted SK-N-SH cells (Fig. 5c-h).
NCAM1 is a glycoprotein found as a cell surface molecule that is known to play a critical role in cell adhesion and is involved in the neurite outgrowth process (Sytnyk et al. 2017). PAX3 binding site is present in NCAM1 promoter and might initiate the NCAM transcription (Chalepakis et al. 1994). The paired domain of PAX3 alone could able to bind to target sequences and inhibit their target gene expression (Chalepakis et al. 1994). Further reports shows that the overexpression of PAX3 increases the PSA-NCAM which is necessary for the axonal growth (Mayanil et al. 2000;Jara et al. 2022). PAX3b is a shorter isoform that consists only of the paired box domain and is able to bind to the paired box recognition sites (Wang et al. 2006;Boudjadi et al. 2018). The observed neuronal outgrowth by functional PAX3 in PAX3bdepleted neuroblastoma cells could be due to NCAM1 activation; however, it needs further validation. This data also suggested that the identification of PAX3 isoform-specific interacting partners will be helpful to understand the function of various PAX3 isoforms in altering their target gene expression. Collectively our experimental evidences suggested that the presence of PAX3b isoform might be responsible for the differentiation defect observed in SK-N-SH cells, and induction of neuronal differentiation in these cells relies on the restoration of functional PAX3 (Fig. 5i).

Restoration of functional PAX3 may have therapeutic potential
PAX3 mutations are widely reported in Waardenburg syndrome in human and splotch phenotype in mice models (Epstein et al. 1993;Boudjadi et al. 2018). Patients with Waardenburg syndrome are reported with neurological defects (Pingault et al. 2010). PAX3 heterozygous mutation resulted in cranial neural tube defects, and homozygous mutation lead to both spinal and cranial neural tube defects (Palmer et al. 2021). Mutations in the DNA-binding domains of PAX3 associated with neural tube defects including spina bifida (Hol et al. 1995;Fleming and Copp 2000) and PAX3 locus deletion have been reported in the autism patients (Borg et al. 2002). Further, the development of cranial nerves has been severely affected in the mice with PAX3 mutation (Mar et al. 2005). The reported PAX3 alterations may be responsible for the development of these neurological disorders. Strikingly, reexpression of PAX3 reported in neuroblastoma is suspected to play a role in their development particularly by inhibiting differentiation; however, how developmentally important gene reexpression causes differentiation defects remains puzzling. Here our results demonstrated that poor neuronal differentiation in neuroblastoma cells are associated with low level/activity of PAX3, and the presence of PAX3b isoform may hamper PAX3 activation in neuroblastoma cells. However, the mechanism behind the PAX3b-mediated inhibition of PAX3 if any needs to be explored. In addition, our unpublished data also revealed that restoration of functional PAX3 inhibits malignant phenotype in PAX3b isoform-depleted tumor cells (Data not shown). Further, we found that the targeting PAX3b or restoration functional PAX3 could be helpful to enhance the neuronal differentiation in neuroblastoma. Overall our findings suggested that functional restoration of PAX3 may have a potential therapeutic strategy for the disorder where PAX3 is involved. Specifically, the identification of pharmacologically active compound/s that could modulate PAX3 function will have the potential to be developed as targeted therapeutics for PAX3-implicated diseases.