Alterations of serine racemase expression determine proliferation and differentiation of neuroblastoma cells

Although the role of serine racemase (SR) in neuropsychiatric disorders has been extensively studied, its role in cell proliferation and differentiation remains unclear. Deletion of Srr, the encoding gene for SR, has been shown to reduce dendritic arborization and dendritic spine density in the brains of adult mice, whereas increased SR levels have been associated with differentiation in cell cultures. Previously, we demonstrated that valproic acid induces differentiation in the N2A neuroblastoma cell line, and that this differentiation is associated with increased SR expression. These observations suggest that SR may have a role in cell proliferation and differentiation. We herein found that both valproic acid and all‐trans retinoic acid induced N2A differentiation. In contrast, knockdown of SR reduced levels of differentiation, increased N2A proliferation, promoted cell cycle entry, and modulated expression of cell cycle‐related proteins. To further evaluate the effects of SR expression on cell proliferation and differentiation, we used an in vivo model of neuroblastoma in nude mice. N2A cells stably expressing scramble shRNA (Srrwt‐N2A) or specific Srr shRNA (Srrkd‐N2A) were subcutaneously injected into nude mice. The weights and volumes of Srrwt‐N2A‐derived tumors were lower than Srrkd‐N2A‐derived tumors. Furthermore, Srrwt‐N2A‐derived tumors were significantly mitigated by intraperitoneal injection of valproic acid, whereas Srrkd‐N2A‐derived tumors were unaffected. Taken together, our findings demonstrate for the first time that alterations in SR expression determine the transition between proliferation and differentiation in neural progenitor cells. Thus, in addition to its well‐established roles in neuropsychiatric disorders, our study has highlighted a novel role for SR in cell proliferation and differentiation.


| INTRODUCTION
Serine racemase (SR), a pyridoxal-5′-phosphate-dependent enzyme, catalyzes the synthesis of D-serine from L-serine through racemization, and the generation of pyruvate and ammonia through an α, βelimination reaction. 1 During the reaction, pyridoxal-5′-phosphate binds to SR and reacts with L-serine to produce an external aldimine intermediate. De-protonation of the aldimine intermediate gives rise to a carbanion intermediate, which produces D-serine upon re-protonation, and pyruvate and ammonia upon elimination of the βhydroxyl group. 2 D-serine, the racemization product, is an endogenous co-agonist of the N-methyl-D-aspartate receptor (NMDA-R) that regulates NMDA-R activity. 3,4 D-serine is involved in essential physiological processes associated with NMDA-R activity including neurotransmission, synaptic plasticity, neurodevelopment, and cell migration. [5][6][7] Thus, alterations in SR expression and D-serine levels are involved in neuropsychiatric disorders linked to abnormalities in NMDA-R activity. Excitotoxicity, mostly due to NMDA-R overactivation, is associated with neurodegeneration in Alzheimer's disease (AD). [8][9][10] Furthermore, overactivation of NMDA-Rs has been associated with increased SR expression in the AD brain. [11][12][13][14] Conversely, hypoactivation of NMDA-Rs is linked to schizophrenia, 15,16 a dysfunction with defects in neuronal migration and differentiation. 17 Consistent with hypoactivation of NMDA-Rs in the schizophrenic brain, decreased levels of SR and D-serine have been reported in the brains of schizophrenic animal models. 18,19 Remarkably, reduced stability of SR in astrocytes has been associated with decreased dendritic arborization in neurons in an astrocyte-neuron co-culture system. 20 In animals, disruption of the binding between SR and Disrupted-In-Schizophrenia-1, a genetic risk factor for schizophrenia, was shown to lead to rapid degradation of SR and the appearance of schizophrenialike behavior. 19 Furthermore, deletion of Srr (Genbank no.27364) in mice was found to reduce dendritic arbor complexity, as well as decrease the spine density of pyramidal neurons in the primary somatosensory cortex and neurons in the hippocampus. 21,22 In contrast, SR expression is increased in differentiated cells or tissues. For example, SR expression is increased in confluent keratinocyte cultures compared to cultures in the growing phase. 23 P19 murine embryonal carcinoma cells are a widely accepted in vitro model for early neurogenesis. 24 P19 cells express SR when induced to differentiate into neurons or glia with all-trans retinoic acid (RA). Interestingly, retinoic acid (RA)-induced differentiation is not blocked by inhibiting D-serine synthesis. 25 In contrast, increased SR expression was found in colorectal adenoma, with more significantly increased levels in proliferative and advanced stage tumors. 26 Thus, taken together, these studies suggest that changes in SR expression may specify neuronal fate, that is, different levels of SR may determine the proliferation or differentiation status in neural progenitor cells.
Neuronal development involves four stages: neurogenesis, migration, differentiation, and neurite outgrowth. Cortical neurogenesis starts as neuroepithelial cells, which transform into radial glial cells (RGCs), giving rise to neuronal precursors. RGCs form scaffolds along which precursor cells migrate into cortical layers and differentiate into different types of neurons in response to intrinsic and extrinsic cues. Finally, functional neurons form due to the growth of dendrites and axons. 27 Interestingly, neuronal precursors acquire their pre-determined fate and differentiate into specific types of neurons in different layers of the cortex. During this transformation, exiting the cell cycle is a prerequisite prior to the initiation of differentiation. Although many studies have focused on this process, little is known about the mechanisms associated with this transition. In eukaryotic cells, cyclin-dependent kinases (CDKs) and their associated cyclins form complexes, which induce cell cycle progression, and which are negatively regulated by CDK inhibitors (CKIs). 28 Of the known CKIs, p27 is critical for neurogenesis. For example, p27 has been implicated in the cell cycle arrest of neural progenitors in the developing brain. 29,30 Specifically, p27 promotes cell cycle arrest in the G1 phase in different cell models. 31,32 During the final stage of neurogenesis, neurite outgrowth, Rho family GTPases regulate actin cytoskeleton dynamics and are involved in neurite development. 33,34 Recently, we found that valproic acid (VPA) induced SR expression under differentiating conditions, but failed to demonstrate for the first time that alterations in SR expression determine the transition between proliferation and differentiation in neural progenitor cells.
Thus, in addition to its well-established roles in neuropsychiatric disorders, our study has highlighted a novel role for SR in cell proliferation and differentiation.

K E Y W O R D S
all-trans retinoic acid, cell cycle, differentiation, D-serine, neuroblastoma, proliferation, serine racemase, valproic acid increase SR expression in undifferentiated milieu. 35 Thus, in the current study, we aimed to determine whether and how SR modulates differentiation of N2A cells. Using in vitro cultures, we found that SR mediated differentiation induced by VPA or RA in N2A cells, as evidenced by neuronal polarization and neurite outgrowth. In contrast, knockdown of SR promoted proliferation and altered expression of cell cyclerelated proteins. To confirm our hypothesis that SR determines N2A in proliferation and differentiation, we generated an in vivo model of neuroblastoma in nude mice. We found that stable knockdown of SR increased tumor proliferation, while VPA treatment mitigated the proliferation of tumors only in cells with normal SR expression levels. In summary, this study is the first to demonstrate that SR is involved in proliferation and differentiation in neural progenitor cells.

| Ethics
The procedures of manipulating animals and of reducing animal discomfort followed ARRIVE guideline (https://arriv eguid elines.org/), and were approved by the Institutional Review Board of Wenzhou Medical University (approval number# wydw2020-0297).

| N2A cell culture
Neuro-2a (N2A) cells were obtained from ATCC. N2A cells were cultured in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), in a humidified 5% CO2 incubator at 37°C. Under differentiated condition, the media were switched into DMEM plus 1% FBS containing VPA or all-trans retinoic acid (20 μM). To establish SR-knockdown N2A cell line, the plasmids expressing srr shRNA under the control of a U6 snRNA promoter in pGPU6/GFP/Neo (GenePharma, Shanghai) were transfected into N2A cells. Basing on knockdown efficacy, stable transfection of the selected plasmid was used for generating SR-knockdown cell line. The transfected N2A cells were selected with fluorescenceactivated cell sorting (FACS), utilizing GFP fluorescence and further being selected with addition of G418.

| Immunofluorescence
Similar to our previous protocol, 36 N2A cells were cultured on polyethyleneimine-coated coverslips sitting in 6-well plates and treated. Before proceeding to immunofluorescence, the cells were washed with PBS and fixed in 4% paraformaldehyde for 5-10 min, and blocked in goat sera for 45 min. After blocking, N2A cells were incubated with βIIItubulin antibody (1,250) in 0.1% Triton-X100 for 2 h, followed by being incubated with FITC-conjugated goat anti-mouse IgG (1, 1000) in 0.1% Triton-X100 for another 45 min. The nuclei were stained with DAPI for 3 min. The images were captured under fluorescence microscope (DMi8, Leica Biosystems).

| Cell transfection
The Srr kd -N2A (sh445) were plated in 6-well plates at the density of 400 000 cells/well, and cultured in DMEM plus 10% FBS. After 24 h in culture, the cells were transiently transfected with 2 μg/well empty vector (EV) or equal amount of SR-expressing plasmid (Com), in DMEM, for 4 h. After transfection, the media were switched into DMEM plus 1% FBS containing 1 mM VPA for 48 h before analysis. Similarly, the Srr wt -N2A were plated in 6-well plates at the density of 400 000 cells/well, and cultured in DMEM plus 10% FBS. After 24 h in culture, the cells were transiently transfected with 2 μg/well empty vector (EV) and equal amount of SR-expressing plasmid to overexpress SR (OE), in DMEM, for 4 h. After the transfection, the media were switched into DMEM plus 1% FBS containing 1 mM VPA for 48 h before analysis.

| Reverse transcription-polymerase chain reaction (RT-PCR)
N2A cells were cultured in 6-well plates at the density of 400 000 cell/well, in DMEM plus 10% FBS. After 24 h, the culture media were switched to DMEM plus 1% FBS containing 1 mM VPA. At different timepoints of treatment, the cells were used for RNA extraction with Trizol reagent. Reverse transcription reaction was conducted to synthesize cDNA by use of a PrimeScript™ RT reagent kit with gDNA eraser (Takara Bio). Twenty nanograms of cDNA was used as a template and the primers used to amplify Srr and actin mRNA were similar as our previous study. 35 The Power SYBR® Green PCR master mixture (Life Technology, NY, USA) was used as the reaction dye for the RT-qPCR reaction and all PCR reactions were conducted in a 96-well ABI plate format in ABI7500 (Applied Biosystems, USA).

| Western blotting
N2A cells were harvested and homogenized in Super RIPA buffer. Proteins were resolved on 12% SDS-PAGE gels and transferred to nitrocellulose membranes. After sequential incubation with primary and secondary antibody, the membranes were developed with chemiluminescence system kit (ThermoFisher). Images of immunoblots were obtained with FlourChem E Systems (Proteinsimple, Santa Clara, MA, USA) and the optic densities of the immunoblots were quantified using ImageJ software (National Institutes of Health, Bethesda, MD, USA).

expressing plasmid
Trizol reagents were used to extract RNA from N2A cell. With the extracted RNAs as a template, cDNA was synthesized with reverse transcription reaction kit (PrimeScript™ RT reagent kit with gDNA eraser). By use of an enzyme mixture consisting of an Ex Taq poly-merase® (cat# RR001A, Takara Bio) and a high-fidelity Taq polymerase (PrimeSTAR® HS DNA polymerase, Takara Bio) at the ratio of 1: 1, PCR reaction was used to amplify Srr cDNA product with the forward primer, 5′-ATGTGTGCTCAGTACT GCATCT-3′, and the reverse primer, 5′-CTAAACAGAAACCGTCTG GTAAGGAG-3′. The PCR products were resolved with DNA electrophoresis and ~1 kbp DNA fragment was purified from agarose gel. With the purified product as a template, PCR reaction was used to synthesize product incorporating desired restriction sites in which the Kozak sequence to enhance translation. The forward primer was 5′-AAGATCTCGAG GCCACCATGTGTG CTCAGTACTGCATCT-3′, with BglII restriction site and the Kozak consensus sequence underlined, respectively. The reverse primer was 5′-CAGAAAC CGTCTGGTAAGGAGTACCCATACGACGTCCCAG ACTACGCTTAACTGCAG1TCTA GA-3′, with PstI and XbaI restriction site underlined, respectively. After PCR reaction, the product was ligated to pMD18-T vector. After screening targeted clone with sequencing, the bacterial clone was amplified and used to extract plasmid. The plasmid was cut by sequential restriction of KpnI and XbaI and ligated to similarly restricted pcDNA3.1. The ligated product was transformed into DH5α Escherichia coli bacteria and the plasmid was extracted from the multiplied bacteria as a SR-expressing plasmid.

N2A cell line
Initially, three plasmids expressing Srr shRNA were transiently transfected into N2A cells, respectively and Srr mRNA and SR protein levels were examined so as to screen knockdown efficacy. The plasmids screened included: sh336, GCTCAGTACTGC ATCTCCTTT; sh445, GGCGCAATCTCTTCTT CAAAT; sh597, GCTCTCACC TATGCT GCTAAA. The expression of each shRNA was driven by a U6 snRNA promoter in the plasmid, pGPU6/ GFP/Neo (GenePharma, Shanghai). In these plasmids, the encoding genes for Srr shRNAs and GFP fluorescence protein were transcribed as independent transcripts. Since interference plasmid sh445 reached maximal effect, stable transfection of the GFP-containing plasmids (sh445) was used for generating SR-knockdown cell line. N2A cells were cultured in 6-well plates at the density of 400 000 cell/well, in DMEM plus 10% FBS, until 60%-70% confluency. The culture media were switched into DMEM and transfected with pGPU6/GFP/Neo (GenePharma, Shanghai), expressing Srr shRNA (sh445, GGCGCA ATCTCTTCTTCA AAT) or scramble RNA (NC, TTCTCCGAACGTGTCACGT) at 2 μg/well. After 4 h, the media were switched to DMEM plus 10% FBS supplemented with penicillin (100 unit/ml) and streptomycin (100 μg/ml) (PS) for another 24 h. The transfected N2A cells were gently trypsinized, suspended, and selected with FACS, basing on GFP fluorescence. The selected N2A cells by FACS were cultured in DMEM plus 10% FBS supplemented with PS and G418 (300 μg/ml) for three days. G418 was added to the culture to exclude N2A cells dislodging the transfected plasmids. The selection by addition of G418 and FACS were repeated at least five times to obtain high-purified SR-knockdown N2A cells.

| Neurite outgrowth assay
The neurons from 10-15 randomly selected fields for each preparation, at least 50 cells in each field images, were captured under LEICA DMi8 fluorescence microscope. All the neurons in the selected fields were counted and quantified. The measurement of neurite length was conducted in triplicate culture preparations and the values were averaged from between 150-200 cells in either group. The neurons were labeled with β-III tubulin and considered differentiated when neurite lengths were twofold more than the soma diameter. 28,34,37 2.11 | Fluorescence-activated cell sorting N2A cells were cultured in 6-well plates at the density of 400 000 cells/well, in DMEM plus 10% FBS. After 24 h, the culture media were switched into DMEM plus 1% FBS and the cells were subject to vehicle or VPA treatment for 48 h. After treatment, the cells were subject to gentle trypsinization and subsequent centrifugation. The supernatants were discarded and the precipitates were washed with PBS. With addition of ice-cold 70% ethanol, the cells were dispersed evenly for storage at −20°C until use. Before cell cycle analysis, the cells were washed and subject to propidium (PI) staining. With a cell cycle and apoptosis analysis kit (C1052, Beyotime Biotechnology), cell cycle was analyzed with BD FACSCalibur basing on PI staining intensity (BD Biosciences, San Jose, CA, USA).

| EdU incorporation assay
5-ethynyl-2′-deoxyuridine (EdU) labeling was used for monitoring N2A proliferation. N2A cells were cultured on glass slides sitting in 12-well plates and treated by VPA (1 mM) in DMEM plus 1% FBS for 48 h. EdU was added at a final concentration,10 μM, for 2 h. N2A cells were fixed in 4% paraformaldehyde at room temperature for 10 min, then washed with PBS three times. After wash, the neurons were blocked with 3% bovine serum albumin and permeabilized with 0.2% Triton-X100 at room temperature for 20 min. After wash, the neurons were incubated with developing solution (Click-iT™ EdU imaging kit) at room temperature for 30 min in the dark. The cells were coverslipped, and images were captured with a fluorescence microscope (LEICA DMi8).

| Establishment of neuroblastoma model with nude mice
Nude mice were purchased from Charles River Laboratories (Beijing) and fed in a standard pathogenfree animal facility. At the age of six weeks, the mice were used for experiment. The modified N2A cells as described above were gently digested by trypsin to single cell suspension and washed. The cells were made into 10 7 cells/ ml cell suspension. By route of subcutaneous injection, 0.1 ml cell suspension was bilaterally injected into nude mice at the junction of hindlimb and abdomen. After 3.5 weeks, the mice developing approximately equal sizes of tumors were evenly assigned to groups either subject to intraperitoneal injection of vehicle or VPA (250 mg/ kg). After two weeks, the mice were euthanized and tumors were immediately excised, using for evaluating the weights and sizes. The isolated tissues from tumor were also used for qRT-PCR and immunofluorescence.

| R2 program analysis
The program of R2:Genomics Analysis and Visualization Platform (http://r2.amc.nl) together with Kaplan scanning were used to predict the correlation between SR expression and the prognosis of neuroblastoma patients. Neuroblastoma Oberthuer (ArrayExpress:E-TABM-38), Neuroblastoma Versteeg (GEO:GSE16476 88/122), and Neuroblastoma SEQC (GEO:GSE62564), publicly available datasets, were used for the analysis. In brief, R2 program calculated the optimal cutoff expression for Srr by dividing the patients in a good and bad prognosis cohort. According to the expression of Srr, samples within a dataset were sorted and divided into two groups, high and low level of expression groups, on the basis of a cutoff expression value. All cutoff expression levels and their resulting groups were analyzed for survival. The cutoff level was reported and used to generate Kaplan-Meier graphs.

| Quantification and statistics
Quantification of neurites was conducted under blinded manner, that is, the investigator conducting measurement was blinded to treatment condition. The images of neurons were chosen from 9-15 randomly selected fields in each preparation, at least 50 cells in each field. Quantification of the weights and the volumes were also conducted under blinded mode. The tumor weights were evaluated by use of an analytic balance. The tumor volumes were measured with calipers and calculated with the formula(A 2 XB)/2, where A is the shorter diameter and B is the longer diameter, similar as previous protocol. 38,39 All the data were presented as means ± SEM unless stated. Data were examined with Shapiro-Wilk test for normality distribution. For parametric data, Student's t-test was used to compare the differences between two groups. In experiments with more than two groups, the data were subject to one-way ANOVA, followed by Tukey's post-hoc test. When p < .05, differences were regarded as significance. Survival rates were estimated by the Kaplan-Meier method, and the log-rank test was used to assess survival difference.

| Elevated SR expression is associated with increased neuronal differentiation
Previously, we demonstrated that VPA increased SR expression in cells cultured under low serum conditions (Dulbecco's Modified Eagle Medium [DMEM] containing 1% fetal bovine serum [FBS]), but not in high serum conditions (DMEM containing 10% FBS) compared with the sham treatment. 35 Staining cells with the neuronal marker, βIIItubulin, allows differentiated neurons to be identified by the presence of growing neurites, which are at least 1.5-fold longer than the soma diameter as previously reported. 28,34,37 In this study, the neurons containing neurites at least two-fold longer than the neuronal soma were considered to be differentiated neurons ( Figure 1A). N2A cells did not differentiate when cultured in DMEM containing 10% FBS. 35 However, in DMEM containing 1% FBS, a small number of N2A cells were found to differentiate. For example, after 24, 48, 72, and 96 h, the percentage of differentiating N2A cells was 1.78% (p = .005 vs. 0 h), 2.71% (p = .001), 5.92% (p < .001), and 10.38% (p < .001), respectively, compared with the 0 h time point (Supporting Information Figure S1A,B). Consistent with the relatively higher levels of differentiation at 72 and 96 h, we found that Srr mRNA levels increased to 160% (p = .011) and 206% (p = .001), respectively, compared with levels at 0 h (Supporting Information Figure S1C Figure 1B). We also found that VPA treatment led to increased Srr mRNA expression levels of 183% (p = .011), 330% (p = .001), 477% (p < .011), and 681% (p = .011) at 24, 48, 72, and 96 h, respectively, compared with levels observed at 0 h ( Figure 1C). Thus, our findings suggested that VPA induced differentiation of N2A cells, with a concomitant increase in Srr mRNA expression. However, the role of SR expression in neuronal differentiation remains unclear.
We next examined the effects of SR knockdown on neuronal differentiation using shRNA. Srr mRNA and SR protein levels were measured after transient transfection of plasmids expressing Srr shRNA (sh336, sh445, and sh597). Srr mRNA expression levels were decreased to 44.8% (p = .002), 36.9% (p = .006), and 40.9% (p = .006), and SR protein levels were reduced to 48.68% (p = .001), 43.98% (p = .001), and 64.97% (p = .007) after transfection with sh336, sh445, and sh597, respectively, compared with control plasmids expressing scrambled shRNA (Supporting Information Figure S2A-C). Since the plasmid sh445 resulted in the largest decrease in Srr mRNA (63.1%) and SR protein (56.02%) expression levels in DMEM plus 10% FBS (Supporting Information Figure S2A-C), we generated a stable SR-knockdown cell line (namely, Srr kd -N2A) using sh445. Cells stably expressing scrambled shRNA were used as a control cell line, that is, Srr wt -N2A (Supporting Information Figure S2D Figure 2A,B). A higher percentage of Srr wt -N2A cells underwent differentiation (Figure 2A,B), similar to the above findings ( Figure 1A,B). Similarly, RA induced N2A differentiation, with longer treatment times leading to an increased percentage of differentiating cells (Supporting Information Figure S3A,B). SR expression was also found to gradually increase with a longer treatment time and with an increased level of differentiation (Supporting Information Figure S3C-E). Similar to VPA treatment, RA induced higher percentages of differentiation in Srr wt -N2A cells compared with Srr kd -N2A cells (Supporting Information Figure S3F,G).
The lower levels of differentiation observed in Srr kd -N2A cells were rescued by SR overexpression ( Figure 3A-D). For example, differentiation was induced in 45.99 ± 2.85% of Srr kd -N2A cells overexpressing SR compared with 10.03 ± 1.58% of Srr kd -N2A cells transfected with the control plasmid under VPA induction conditions (p < .0001; Figure 3A-D). Interestingly, addition of Dserine (20 μM) did not rescue the differentiation deficit in Srr kd -N2A cells. For example, 20 μM D-serine-treated cultures contained 9.58 ± 1.50% of differentiated Srr kd -N2A cells compared with 9.27 ± 0.74% in sham-treated cells (p = .755; Supporting Information Figure S4). Notably, overexpression of SR in Srr wt -N2A cells did not promote differentiation ( Figure 3E-H), that is, differentiation was induced in 21.63 ± 1.90% of Srr wt -N2A cells overexpressing SR compared with 23.38 ± 2.04% in Srr wt -N2A cells transfected with the control plasmid (p = .254; Figure 3E-H). F I G U R E 1 VPA induced N2A differentiation. A, N2A cells were cultured on pre-coated coverslip sitting in 12-well plate at the density of 80 000 cells/well. After 24 h, the culture media were switched into differentiated culture condition containing 1 mM VPA. The cells were labeled with βIIItubulin by immunofluorescence at different timepoints (24-96 h) of treatment. The images from triplicate culture preparations were captured under fluorescence microscope and the typical images were shown. Scale bar, 50 μm. (B) The percentages of differentiated cells were calculated as the ratios between differentiated neurons and DAPI-staining cells in the same field. The percentage of differentiated neurons were obtained from 15 randomly selected fields for each preparation, at least 50 cells in each field. ***p < .0001 indicated the differences between treatments versus non-treatment, respectively. One-way ANOVA was used to compare the differences. (C) Srr mRNA levels were determined with RT-qPCR from the cultures treated by VPA. The levels of Srr mRNA at 0 h were set as 1 and the levels at different timepoints were normalized accordingly. **p < .05, ***p < .001 versus 0 h, respectively. The results were averaged from triplicate culture preparations. One-way ANOVA was used to compare the differences.

| Knockdown of SR modulates expression of cell-cycle related proteins
We next analyzed the expression of cell cycle-related proteins. Consistent with our EdU data, we found that the cell proliferation markers, pH3 and Ki-67, were increased in VPA-treated Srr kd -N2A cells compared to VPA-treated Srr wt -N2A cells (p = .005 for pH3 comparison; p = .029 for Ki-67 comparison; Figure 5E,G). In addition, we found that the expression levels of Cyclin D1 and pRb, proteins associated with promoting the G1/S transition, were increased in VPA-treated Srr kd -N2A cells compared to VPA-treated Srr wt -N2A cells (p = .003 for Cyclin D1 comparison; p = .0032 for pRb comparison; Figure 5B,F). In contrast, P21 and P27, proteins that block the G1/S transition, were decreased in VPA-treated Srr kd -N2A cells compared to VPAtreated Srr wt -N2A cells (p < .0001 for P21 comparison; p = .003 for P27 comparison; Figure 5C,D). Under high serum conditions (DMEM plus 10% FBS), no significant differences in Cyclin D1, P21, P27, pH3, pRb, and Ki-67 protein expression levels between Srr kd and Srr WT N2A cells were observed (Supporting Information Figure S5).

| High SR expression is associated with a better prognosis for neuroblastoma patients
To further elucidate the role of SR in cell proliferation and differentiation and its implications in clinical translation, the correlation between SR expression and the prognosis for neuroblastoma patients was predicted with bioinformatic tools, R2: Genomics Analysis and Visualization Platform program (http://r2.amc.nl) Kaplan scanning. We found that high expression of SR was associated with good patient prognosis, whereas low SR expression was correlated with a low probability of survival (Figure 7). Patients with high SR expression levels had a greater than 95% probability of survival during the 24-168 month follow-up, whereas patients with low SR expression levels had ~60% probability of survival during the 24 month follow-up. During the 48-216 month follow-up period, the survival rate of patients expressing low levels of SR was significantly reduced to ~55% (Figure 7).

| DISCUSSION
In the current study, we demonstrated that in vitro, induced expression of SR mediated the VPA-or RA-induced differentiation of neuroblastoma cells, whereas knockdown of SR increased proliferation. These effects were confirmed in an in vivo neuroblastoma model in nude mice. VPA treatment was shown to reduce the size and weight of Srr wt -N2A-derived neuroblastomas, but did not have an effect on Srr kd -N2A-derived tumors compared to vehicle-treated mice. Finally, bioinformatics analysis revealed that high SR expression was linked with high survival rates in neuroblastoma patients, whereas low SR expression levels were associated with low survival rates.
SR has been associated with neuropsychiatric disorders such as AD, [11][12][13][14]41 amyotrophic lateral sclerosis, 42,43 and schizophrenia, 18,19 due to regulation of NMDA-R activity by D-Serine. Here, we demonstrated for the first time that SR expression levels determined whether N2A cells underwent proliferation or differentiation. Our findings indicate that the role of SR in cell proliferation and differentiation was unrelated to D-serine, because overexpression of SR in Srr wt -N2A cells did not promote differentiation, and addition of D-serine did not rescue the inability of Srr kd -N2A cells to differentiate. The correlation between elevated SR expression levels and differentiation observed here was consistent with previous reports. 23,25 In particular, the differentiation of P19 cells was previously shown to be coupled with SR expression, but was not affected by the inhibition of D-serine synthesis. 25 Similar to these findings, in our study we demonstrated that addition of D-serine did not rescue the differentiation deficit in SR-knockdown N2A cells. Furthermore, we showed that modulation The images were acquired under fluorescence microscope. The images were typical of quadruplicate culture preparations. Low panel, the cells positive for EdU staining were counted and the proliferation was represented as the ratios between the numbers of EdU + cells and the numbers of DAPI + cells. The ratios were averaged from 9-15 randomly selected fields in each culture preparations. **p = .005. of cell cycle-related proteins accounted for the differentiation deficit in SR-knockdown cells. Whether cells undergo proliferation or differentiation depends on the activity of cell cycle-promoting proteins such as cyclin-CDK complexes, as well as cell cycle-exit proteins such as P27 (Kip1) and P21 (Cip1). In neuroblastoma cells, P27 (Kip1) has an important role in promoting cell cycle exit and differentiation through its interactions with cyclin-CDK complexes and inhibition of the kinase activity. In addition, P27 stabilizes NG2, a transcriptional factor, thereby promoting differentiation. 44 Compared with Srr wt -N2A cells, Srr kd -N2A cells expressed lower levels of P27 Kip1 but higher levels of cyclin D1 under the induction of VPA. These expression patterns suggest that cell cycle exit is dependent on the regulation of cell cycle-related proteins under conditions that lead to increased SR expression. However, the mechanism by which SR modulates the expression of cell cycle-related proteins remains unclear.
Since increased SR expression promotes differentiation of N2A cells, this could provide a novel strategy for treating neuroblastoma. Neuroblastoma, originating F I G U R E 5 Stable knockdown of SR modulated expression of cell cycle-related proteins. The Srr kd -N2A (sh445) or Srr wt N2A neurons (NC) was cultured in 6-well plates at the density of 400 000 cells/well for 24 h. The cultures were switched into differentiated condition containing 1 mM VPA for 48 h. The cells were harvested for immunoblotting. Knockdown of SR in Srr kd (sh445)-N2A decreased SR, ***p < .0001 (A); increased CyclinD1, **p = .003 (B); decreased P21, ***p < .0001 (C); decreased P27, **p = .003 (D); increased pH3, **p = .005 (E); increased pRB, **p = .0032 (F); increased Ki67, *p = .029 (G), relative to those in Srr wt N2A neurons. Student's t-test was used to compare the differences. from the sympathoadrenal nervous system, is an extracranial tumor which accounts for more than 7% of malignancies in pediatric patients and ~15% of all pediatric deaths due to cancers. 45 Although overall survival has improved, ~36% of patients have poor prognosis due to metastasis and undifferentiation-related malignancy. 46 For these pediatric patients, overall survival rates remain below 40% despite multimodal therapies including radiotherapy, myeloablative chemotherapy, immunotherapy, and aggressive surgery. 47 Although the detailed molecular events driving neuroblastoma are unclear, the genes regulating the proliferation and differentiation of neural crest cells are thought to be involved in the initiation and progression of neuroblastoma. 48 For example, anaplastic lymphoma kinase mutation or amplification of MYCN are associated with the occurrence of neuroblastoma, because these mutations cause uncontrolled cell proliferation while blocking differentiation. 49,50 Thus, studies focusing on neuronal fate determination and examining the factors that contribute to the proliferation and differentiation of neuroblastoma cells are important for addressing the challenges of neuroblastoma therapy. Induction of differentiation has been shown to be effective in neuroblastoma therapy. For example, the application of histone deacetylase inhibitors has emerged as a potential strategy to treat neuroblastoma, because these inhibitors possess the ability to inhibit tumor proliferation, or induce differentiation, apoptosis and cell cycle arrest in the tumors. [51][52][53][54] Our findings suggested that SR mediates VPA-or RA-induced differentiation in N2A cells. Thus, SR may be a novel molecular target for treating neuroblastoma.
In summary, this study demonstrated that SR expression determined whether N2A cells underwent proliferation or differentiation. Furthermore, the proliferation and differentiation was associated with the modulation of cell F I G U R E 6 Stable knockdown of SR promoted tumor growth and VPA mitigated the growth in Srr wt -N2A-derived tumors but not in Srr kd -N2A(sh445)-derived tumors. Establishment of neuroblastoma model with nude mice was conducted as described in the materials and methods. The tumors were excised immediately after euthanization and the volumes were measured with a caliper (A) and the weights were evaluated by use of an analytic balance (B). (A) The volumes of the sh445-derived tumors (sh445, 2.15 ± 0.37 cm 3 ) were bigger than those of the Srr wt -derived tumors (NC, 1.45 ± 0.40 cm 3 ) (*p = .02). VPA injection reduced tumor sizes in NC (NC VPA, 0.68 ± 0.23 cm 3 ) compared with PBS injection (NC) (***p < .0001) whereas VPA injection did not reduced tumor sizes in sh445 mice (sh445 VPA, 2.18 ± 0.38 cm 3 ) compared with PBS injection (sh445) (p = .561). (B) The values of the weights of the sh445-derived tumors (sh445, 1.24 ± 0.2 g) were bigger than Srr wt -derived tumors(NC, 0.98 ± 0.1 g) (*p = .03). VPA injection reduced tumor in NC(NC VPA, 0.42 ± 0.11 g) compared with PBS injection (NC) (***p < .0001) whereas VPA injection did not reduce tumor in sh445 mice (sh445 VPA, 1.32 ± 0.24 g) compared with PBS injection (sh445) (p = .25). The values were averaged from tumors isolated from five mice in each group. One-way ANOVA was used to compare the differences. (C) The excised tissues from tumors were also subject to RT-qPCR to evaluate the levels of Srr mRNA. Srr mRNA levels in sh445-derived tumor were lower than NC (**p = .002). VPA injection increased Srr mRNA levels in NC (**p = .001) or in sh445derived tumors ( # p = .015), relative to PBS injection, respectively. The results were averaged from tissues isolated from three mice. Oneway ANOVA was used to compare the differences. (D) The tumors isolated from PBS-injection Srr wt N2A-derived neuroblastoma mice (NC),VPA-injection Srr wt N2A-derived neuroblastoma mice (NC VPA), PBS-injection sh445 N2A-derived neuroblastoma mice (sh445), and VPA-injection sh445 N2A-derived neuroblastoma mice (sh445 VPA) were displayed. (E) The isolated tissues from tumors were made into paraffin-embedded sections and were immunoblotted against SR for the tissues. DAPI was used to indicate nuclei and SR staining and DAPI staining were merged in the last column. Omitting the primary antibody did not produce staining. Scale bar, 50 μm.
cycle-related protein expression. Our study highlights a novel role for SR in cell proliferation and differentiation.

AUTHOR CONTRIBUTIONS
He Zhang performed most experiments, collected and analyzed the data. Jinfang Lu constructed SR-expressing plasmid, cell culture, and western blot. Huiping Shang participated in cloning plasmid. Juan Chen conducted FACS analysis. Zhengxiu Lin participated in tumor histology. Yimei Liu participated in western blot. Xianwei Wang participated in FACS analysis. Liping Song participated in establishing SR-knockdown N2A cell and cell culture. Xue Jiang and Haiyan Jiang established neuroblastoma model with nude mice. Jiandong Shi and Dongsheng Yan helped with data mining. Shengzhou Wu conceptualized the project, analyzed data, and wrote the manuscript.

F I G U R E 7
The potential correlation between SR expression and the prognosis of neuroblastoma patients. R2 program as described in the materials and methods was used to predict the correlation between SR expression and the prognosis of neuroblastoma patients. High level of SR expression (n = 22) was associated with good prognosis in neuroblastoma patients whereas low level of SR expression (n = 66) with bad prognosis in neuroblastoma patients. Survival rates were estimated by the Kaplan-Meier method, and the log-rank test was used to assess survival difference. **p = .0016.