An ALYREF-MYCN coactivator complex drives neuroblastoma tumorigenesis through effects on USP3 and MYCN stability

To achieve the very high oncoprotein levels required to drive the malignant state cancer cells utilise the ubiquitin proteasome system to upregulate transcription factor levels. Here our analyses identify ALYREF, expressed from the most common genetic copy number variation in neuroblastoma, chromosome 17q21-ter gain as a key regulator of MYCN protein turnover. We show strong co-operativity between ALYREF and MYCN from transgenic models of neuroblastoma in vitro and in vivo. The two proteins form a nuclear coactivator complex which stimulates transcription of the ubiquitin specific peptidase 3, USP3. We show that increased USP3 levels reduce K-48- and K-63-linked ubiquitination of MYCN, thus driving up MYCN protein stability. In the MYCN-ALYREF-USP3 signal, ALYREF is required for MYCN effects on the malignant phenotype and that of USP3 on MYCN stability. This data defines a MYCN oncoprotein dependency state which provides a rationale for future pharmacological studies. Neuroblastoma (NB) is often driven by MYCN amplification. Here, the authors show that the most frequent genetic lesion, gain of 17q21-ter in NB leads to overexpression of ALYREF, which forms a complex with MYCN, regulating MYCN stability via the deubiquitinating enzyme, USP3.

M YCN is a short-lived transcription factor, tightly controlled by the ubiquitin-proteasome system. Oncogenic stabilization of MYCN promotes target gene recognition and transcriptional programs involved in almost every aspect of tumorigenesis 1 . Amplification of the MYCN oncogene from 2p24 occurs in 25% of neuroblastoma patients, and is a wellrecognized marker of tumor aggressiveness 2,3 . MYCN-amplification in primary neuroblastoma tissues shows strong statistical association with 17q21-ter gain 4,5 , the most frequent genetic aberration in neuroblastoma at an incidence of 38-65% 4,6-11 . Genes identified within this chromosomal region such as insulin growth factor 2 binding protein 1 12 , survivin 13 , and JMJD6 14 have been reported to promote neuroblastoma cell survival through a wide range of mechanisms, however, the mechanistic relationship between MYCN and 17q21-ter is still not understood. As there are no specific inhibitors currently available for high-risk neuroblastoma patients with these features, there is an urgent need to identify new targetable molecular vulnerabilities.
The ALYREF gene is sited at chromosome 17q25.3, and codes for a ubiquitously expressed nuclear chaperone protein that controls many biological processes. ALYREF is a bZIP enhancing factor 15 that regulates DNA binding and transcriptional activity of RUNX1B and c-Myb 16 , LEF-1 and AML 17 , and E2F2 18 . ALYREF can bind to RNA polymerase II to assist in transcriptional elongation 19 . ALYREF is also a component of the TREX protein complex regulating nuclear export of mRNAs 20 and splicing 21 . ALYREF expression is dysregulated in several human cancers 22,23 . Suppression of ALYREF expression resulted in decreased cell proliferation 24 , and the migratory capacity of oral squamous carcinoma cells 23 . Despite considerable study, it is not yet clear how the expression of ALYREF is increased or how elevated levels of ALYREF promote tumorigenesis.
In this study, we describe ALYREF as a key regulator of MYCN turnover and neuroblastoma tumorigenesis. Bioinformatic analyses demonstrated that elevated ALYREF expression associated with 17q21-ter gain, MYCN-amplification and was a predictor of poor patient survival. Mechanistic experiments revealed that MYCN was dependent on ALYREF to drive neuroblastoma cell proliferation and the two proteins acted in a positive interconnected regulatory pathway that was essential for sustaining mutual, oncogenic expression levels. ALYREF contributed to neuroblastoma cell proliferation by blocking MYCN degradation through direct transcriptional upregulation of USP3, a deubiquitinating enzyme. Genetic suppression of ALYREF expression resulted in reduced neuroblastoma cell proliferation in vitro, and delayed neuroblastoma growth in mice. These findings provide evidence for future pharmacological studies aimed at the development of ALYREF inhibitors as selective therapeutics for neuroblastoma with 17q21-ter gain and MYCN-amplification.

Results
High ALYREF expression associates with 17q21-ter gain, MYCN-amplification, and poor patient prognosis in neuroblastoma. To identify molecular vulnerabilities in the context of 17q21-ter gains we first used a bioinformatic approach and analyzed the US TARGET neuroblastoma tumor cohort of 135 patients with Whole Genome Sequencing (WGS) data 25 . Analysis of the dataset showed a high frequency of gain beyond 17q21.31 (Fig. 1a) in neuroblastoma tissues. Kaplan-Meier analysis of overall survival probability of patients in the same tumor cohort (n = 135) showed that 17q21-ter gain (gain subgroup) was strongly predictive of poor patient survival when compared to patients without 17q21-ter gain (diploid subgroup; Fig. 1b). Amplification of the MYCN oncogene and consequent overexpression of the MYCN oncoprotein occur in 25% of neuroblastoma patients and correlate with poor patient survival 2,3 . Further stratification of Kaplan-Meier survival analysis into four subgroups based on both 17q21-ter gain (gain or diploid) and MYCN-amplification status (amplified; MA, nonamplified; MNA) showed that patients with 17q21-ter gain and MYCN-amplification (gain-MA) had the worst survival probability among the four subgroups with significantly lower survival rates when compared to 17q21-ter gain (gain-MNA) alone or without either aberration (diploid-MNA) (Fig. 1c). We further utilized RNA-seq data from the same TARGET neuroblastoma cohort to analyze 1044 transcripts from the 17q21-ter locus for all of the following five criteria (Fig. 1d): differential gene expression in (1) 17q21-ter-gain patients (Supplementary Data 1) and (2) MYCN-amplified patients (Supplementary Table 1 Table 2), as well as (5) association with poor neuroblastoma patient outcome (Supplementary Table 3). ALYREF was the only gene to pass all five criteria among the 1044 17q21-ter genes (Fig. 1e). Following transcriptome-wide differential gene expression analyses we observed that ALYREF was differentially expressed in patients with either 17q21-ter gain (Fig. 1f) or MYCN-amplification (Fig. 1g). We also observed genes previously associated with 17q21-ter gain in neuroblastoma, such as TBX2 26 , PPMD1 27 , JMJD6 14 , and BIRC5 28 being differentially expressed in 17q21-ter gain patients ( Supplementary Fig. 1a), whereas MYCN mRNA expression associated with MYCN-amplification (Supplementary Fig. 1b) but not 17q21-ter gain ( Supplementary Fig. 1a). Cox proportional hazards (CoxPH) and Kaplan-Meier analyses of overall survival probability of patients in the TARGET cohort (n = 154) showed that high ALYREF expression was strongly predictive of poor patient survival (Fig. 1h). We obtained similar results from the larger RNA-seq SEQC (n = 498) neuroblastoma patient cohort [29][30][31][32] (Supplementary Fig. 1c). High ALYREF expression in this cohort also correlated with other predictors of poor prognosis in the disease, such as (1) MYCN-amplification, (2) advanced clinical stage (3/4), and (3) older age of the patient at diagnosis (Supplementary Fig. 1d). Multivariate CoxPH analysis for both overall and event-free ( Supplementary Fig. 1e) survival in the SEQC cohort (n = 498) showed that high ALYREF expression retained independent prognostic significance, when compared with other prognostic factors such as patient age, clinical stage, and MYCN-amplification. Further additive CoxPH regression analysis for both overall and event-free (Supplementary Fig. 1f) survival showed that clinical stage and MYCNamplification had the greatest impact on the prognostic significance of ALYREF expression. These data suggest that high ALYREF expression strongly predicts poor patient prognosis and that MYCN and ALYREF may co-operate as tumorigenic factors in neuroblastoma. We saw significant positive correlations between ALYREF and MYCN expression levels (Fig. 1i), as well as ALYREF copy number (Fig. 1j), further supporting ALYREF as a key intermediary in the cooperation between 17q21-ter gain and MYCN-amplification in the malignant neuroblastoma phenotype. By further examining pan-cancer cell line gene expression (RNAseq) and copy number profiles using Cancer Cell Line Encyclopedia (CCLE)/Cancer Dependency Map 33 , we observed that neuroblastoma cell lines had the highest ALYREF gene expression and copy number compared with all other cancer types, supporting the principle that ALYREF is an abundant gene target in neuroblastoma ( Supplementary Fig. 1g, h).
MYCN increases neuroblastoma cell viability in an ALYREFdependent manner. To investigate the functional relationship between MYCN and ALYREF in vitro, we evaluated the phenotypic effect of ALYREF knockdown on human SHEP-MYCN3 neuroblastoma cells, which permits doxycyclineinducible overexpression of stably incorporated MYCN, and are therefore ideal for MYCN gene dependency experiments. SHEPMYCN3 cells were transfected with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2 ( Fig. 1k) for 72 h with (lanes 4-6) and without doxycycline (lanes 1-3). As expected, transfection of ALYREF siRNA in MYCN-low and ALYREF-low expressing conditions did not reduce cell viability (Fig. 1l) or proliferation (Fig. 1m, n and Supplementary Fig. 1i). However, MYCN-induced increases in viability (Fig. 1l) and proliferation ( Fig. 1m-n and Supplementary Fig. 1i) were blocked by ALYREF knockdown. Taken together these data showed that MYCNinduced increases in neuroblastoma cell viability and proliferation  were, in part, dependent on ALYREF. ALYREF is known to regulate cell proliferation 23,24 , however, its role in regulating cell survival is unknown. To determine whether ALYREF knockdown affects MYCN-induced increases in cell survival, we investigated the effect of ALYREF inhibition on cell death in doxycyclineinduced (DOX) or not induced (Vehicle) SHEPMYCN3 cells, transfected with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2 ( Supplementary Fig. 1j-k). Abolition of mitochondrial membrane potential is a marker of apoptosis. Mitochondrial membrane potential was measured using a MitoProbe DilC1(5) Assay ( Supplementary Fig. 1j). MYCN overexpression induced by doxycycline treatment did not affect mitochondrial membrane potential ( Supplementary Fig. 1j) at the earlier time points. Only at 48 h was there a slight decrease in mitochondrial membrane potential induced by MYCN overexpression. ALYREF knockdown did not significantly affect the survival of the doxycyclineinduced (DOX) or not induced (Vehicle) SHEPMYCN3 cells. To further investigate necrotic processes consequent on altered ALYREF levels, we assessed the effects of ALYREF knockdown on MYCN-induced survival by measuring SYTOX Green accumulation in the cells ( Supplementary Fig. 1k). The dye is able to penetrate (and then bind to the nucleic acids) only necrotic cells with ruptured plasma membranes, whereas healthy cells with intact surface membranes show significantly lower SYTOX Green staining. As expected, doxycycline-induced MYCN overexpression protected the cells from cell death and significantly decreased the number of necrotic SHEPMYCN3 cells (Supplementary Fig. 1k). However, ALYREF knockdown did not further decrease SYTOX Green accumulation in the cells. Therefore, we conclude that ALYREF predominantly causes induction of proliferation in MYCN overexpressing cells, but does not assist MYCN in regulating cell survival. We also investigated the phenotypic effect of ALYREF knockdown on several human neuroblastoma cell lines with high ALYREF expression ( Supplementary  Fig. 1l-m). Endogenous knockdown of ALYREF in all neuroblastoma cells ( Supplementary Fig. 1l) was accompanied by a significant decrease in cell viability ( Supplementary Fig. 1m), as measured by Alamar Blue assay.
MYCN directly upregulates ALYREF transcription. We next investigated the regulatory relationship between MYCN and ALYREF. High ALYREF mRNA expression correlated with MYCN mRNA expression in the TARGET 25  We next sought evidence that ALYREF mRNA expression in the organ of tumor origin, sympathetic ganglia, correlated with an MYC expression signature from tumor initiation to progression in tissues from the TH-MYCN +/+ transgenic neuroblastoma mouse model 35 . Microarray mRNA expression analysis showed increasing ALYREF and MYC-signature gene expression throughout tumor progression in ganglia isolated from TH-MYCN +/+ mice in comparison to ganglia from wild-type littermates (Fig. 2d). Importantly, high ALYREF mRNA expression strongly correlated with MYCN expression in ganglia tissues throughout tumor progression (Fig. 2e). Analysis of the ALYREF gene promoter sequence and published chromatin-immunoprecipitation (ChIP)sequencing data 36 revealed non-canonical MYCN binding sites CACCTG (at −549 bp) and CAGCTG (+635 bp) in close proximity to the ALYREF transcription start site (TSS, Fig. 2f). We performed ChIP assays with an anti-MYCN antibody or control mouse IgG antibody (Santa Cruz Biotechnologies) followed by real-time PCR with primers targeting a negative control (Control; 2000 bp upstream TSS) and the MYCN peak summit (Promoter, −248 bp) upstream of the ALYREF TSS (Fig. 2g). ChIP assays showed significant MYCN binding 14,37,38 at its putative binding site 248 bp upstream ALYREF TSS. The MYCN antibody immunoprecipitated the ALYREF region 3-fold higher than the negative control region in SK-N-BE(2)C cells and 6-fold in Kelly cells (Fig. 2h), confirming that MYCN directly binds the ALYREF gene promoter. Transfection of MYCN-amplified SK-N-BE(2)C, Kelly (Fig. 2i-j) and CHP-134 cells ( Supplementary   Fig. 1 High ALYREF expression associates with 17q21-ter gain, MYCN-amplification, and poor patient prognosis in neuroblastoma. a Frequency of segmental copy number calls (loss, diploid, gain) spanning Chromosome 17 (10 kilobase genomic bins), from whole-genome sequencing (WGS) of the TARGET cohort (n = 135) 25 . 17q21-ter region is annotated by a black bar, genomic coordinates are provided in megabases. Centromeric regions were excluded (p/q11.1). Overall survival based on b 17q21-ter stratification and c 17q21-ter (gain or diploid) and MYCN status (Amplified; MA, Non-Amplified; MNA) in the TARGET cohort (n = 135). Hazard ratio (HR) from a Cox proportional hazards (CoxPH) model. p-values are from a two-sided log-rank test, and adjusted using the Benjamini-Hochberg method. d RNA-seq (n = 154) and WGS (n = 135) data of the TARGET cohort 25 were utilized to analyse1044 genes on the 17q21-ter region. e Scores of the 1044 genes according to genomic location, genomic distances in megabases (Mb). ALYREF expression between f diploid (n = 50) and 17q21-ter gain (n = 40) and g MYCN non-amplified (n = 121) and MYCN-amplified (n = 33) patients adjusted p-value is from two-sided Wald chi-squared tests, followed by Benjamini-Hochberg method in f and g. h Overall survival of the TARGET cohort (n = 154) dichotomized by median ALYREF expression (RNA-Seq). Hazard ratios (HR) and two-sided log-rank p-values are from a univariate CoxPH model. Correlation between i MYCN and ALYREF expression RNA-seq (n = 154), and j ALYREF expression RNA-seq and ALYREF copy number WGS (n = 90), MYCN-amplified patients are annotated in red. p-values from two-sided Pearson's correlation test in i and j. k Immunoblot for MYCN and ALYREF following ALYREF knockdown in doxycycline-induced (+) vs non-induced (−) SHEPMYCN3 cells. l Cell viability (n = 3), m proliferation (n = 3), and n colony formation analysis of DOX vs Vehicle-treated SHEPMYCN3 cells transfected with ALYREF siRNAs. p-values from two-sided Student's t-test in l and m.  ALYREF participates in a positive interconnected regulatory signaling axis with MYCN. We next assessed whether ALYREF acted in a forward feedback expression loop with MYCN, as we have identified for other MYCN transactivation targets [37][38][39][40] . Silencing endogenous ALYREF with ALYREF siRNA-1 or ALYREF siRNA-2 (lanes 2-3 and 5-6, Fig. 3a) in     Supplementary Fig. 3c). These data suggest that ALYREF-regulated MYCN expression through a posttranslational mechanism. MYCN goes through rapid degradation by the ubiquitin-proteasome system 41 . To examine the hypothesis that ALYREF-regulated MYCN half-life through ubiquitination in neuroblastoma cells, we transfected Kelly and SK-N-BE(2)C cells with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2. Following treatment with the proteasome inhibitor, MG132, the total cellular protein was immunoprecipitated for endogenous MYCN 42 and then probed with an antibody specific for K-48-linked ubiquitin chains, the trigger signal for proteasomal degradation (Fig. 3e). The results showed that ALYREF knockdown increased K-48-linked ubiquitination of MYCN (lanes 5-6 vs lane 4; Fig. 3e) in both cell lines. These data indicate that ALYREF maintains a high level of MYCN expression through inhibition of its protein degradation.
ALYREF forms a protein complex with MYCN in the nucleus. Since ALYREF modulated MYCN protein stability, we examined the possibility of their physical interaction. We transfected HEK293T human embryonic kidney cells with a construct expressing HA-tagged MYCN either alone, or one expressing Flag-tagged ALYREF. An MYCN co-immunoprecipitation (co-IP) of total cellular protein with an HA-tag antibody showed that ALYREF-bound MYCN (lane 6; Fig. 4a), which was confirmed in reciprocal ALYREF co-IP experiments with a Flag-tag antibody (lane 6; Fig. 4b). Separation of total cellular protein into nuclear and cytoplasmic fractions by centrifugation allowed us to show that both MYCN and ALYREF were predominantly localized to the nucleus (lanes 3 and 6; Fig. 4c) in MYCN-amplified neuroblastoma cell lines (Kelly and SK-N-BE(2)C). We next examined total cellular protein (lanes 1-3 and 10-12; To map the domain of MYCN, which was critical for the interaction with ALYREF, we used MYCN deletion mutant expression constructs 43 (Fig. 4e). We co-transfected HEK293Tcells with constructs expressing HA-tagged MYCN deletion mutants and a construct expressing Flag-tagged full-length ALYREF. Flag immunoprecipitation assays showed that ALYREF was present in the protein immunoprecipitates containing HA-  Supplementary Fig. 4c). These data demonstrate that a small region of MYCN containing amino acids 281-345 was crucial for the binding to ALYREF.
ALYREF and MYCN form a transcriptional activator complex which upregulates USP3 expression. ALYREF has been suggested to act as a regulator of DNA binding and modulate the activity of transcription factors [15][16][17][18] . The binding of ALYREF and MYCN proteins in the nucleus, as well as the positive forward feedback expression loop between the two proteins, led us to hypothesize that ALYREF and MYCN cooperated and played a role in upregulating transcriptional programs involved in the ubiquitination process regulating MYCN turnover. To identify gene targets, we first mapped the genome-wide profile of ALYREF-chromatin interactions in SK-N-BE2C cells using ChIPseq. We looked for transcriptional target genes of ALYREF based on the following selection criteria (Fig. 5a): (1) presence of MYCN and RNA Pol II (indicating active transcriptional elongation), as well as two proximate histone modifications associated with transcriptional activation (H3K27ac and H3K4me3) at the gene promoter; (2) having a high fold enrichment (FE) for ALYREF binding of more than 5; (3) expression correlation with neuroblastoma patient survival in the SEQC neuroblastoma patient tumor cohort (n = 498); (4) correlation with MYCN expression level in the SEQC neuroblastoma patient tumor cohort (n = 498); and (5) having known biological function relevant to ubiquitination. We found a total of 3598 ALYREF-binding sites (FDR < 0.05; Supplementary Fig. 5a). We identified 194 Fig. 2 MYCN directly upregulates ALYREF transcription. a Scatter plot, with a linear regression fit for 498 neuroblastoma patients from the SEQC cohort for MYCN vs ALYREF gene expression (RNA-Seq, log 2 RPM). Patients with MYCN-amplification, no amplification, or are of unknown status are annotated. A two-sided Pearson's correlation test was used to derive p-values. b Immunoblot and c densitometry analysis (n = 3 per group) for ALYREF expression in a range of MYCN-amplified (SK-N-BE(2)C, Kelly, CHP-134, SK-N-DZ, LAN-1, IMR-32) and MYCN overexpressing (NBLS), low MYC or MYCN expressing (SH-EP, SK-N-FI), cMYC overexpressing neuroblastoma (SH-SY5Y, SK-N-AS) and normal lung fibroblast (WI-38, MRC5) cells. Differences in ALYREF protein expression were compared using two-sided one-way ANOVA. d mRNA expression of MYC-signature 35 genes and ALYREF in ganglia from TH-MYCN +/+ mice, obtained at different postnatal ages (1, 2, and 6 weeks of age; n = 4 per group; microarray, log 2 ). e Scatter plot, with a linear regression fit from the TH-MYCN +/+ cohort for MYC-signature vs ALYREF gene expression (6 weeks of age; microarray, log 2 ). A two-sided Pearson's correlation test was used to derive p-values. f Schematic of the ALYREF promoter in MYCN-amplified neuroblastoma cells detailing canonical (CACGTG) and non-canonical (CANNTG) MYCN binding motifs. g Schematic of the ALYREF (Promoter) and negative control (Control) target sequences used for ChIP-PCR, detailing the MYCN peak summit and its distance from the transcription start site (TSS). h ChIP-PCRassays (n = 3 per group) for the negative control region (Control; 2000 bp downstream of TSS) or the ALYREF promoter containing the MYCN peak summit (Promoter, −248 bp downstream of TSS). Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in expression levels were compared to the negative control (Control). i Immunoblots j and RT-PCR (n = 3 per group) of MYCN and ALYREF expression following MYCN knockdown. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in expression levels were compared to Control siRNA. Pearson correlation coefficient (r) and respective p-values (p) are displayed. Comparisons were not significant unless otherwise noted. Data is representative of three independent experiments with similar results in b and i. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in c, d, h, and j and 4 mice per age group in d.
ALYREF-regulated genes undergoing transcription, among which 57 genes (Supplementary Table 4) had a FE of more than 5 (Fig. 5b). From these 57 genes, we identified the ubiquitin-specific protease 3 gene (USP3) as satisfying all criteria. USP3 had FE = 6.15987 for ALYREF binding and was ranked as number 6 among the 57 genes (Supplementary Table 4) based on fold enrichment stratification. Integrative Genomic Viewer showed that ALYREF bound the USP3 intronic region 6.8 kb from the TSS (Fig. 5c). To confirm the ChIP-seq data, we performed a ChIP assay with an anti-ALYREF antibody or control IgG antibody (IgG) followed by real-time PCR with primers targeting a negative control region (Control; 2000 bp downstream TSS) or the putative  (Fig. 5d). The antibody against MYCN efficiently immunoprecipitated the intronic region of USP3 carrying the ALYREF-binding sites at 6.8 kb from the TSS and upstream from USP3 exon 1 (Fig. 5e). Finally, double ChIP for an MYCN-ALYREF complex immunoprecipitated the ALYREF-bound intronic site of the USP3 gene some 3-fold higher than the negative control region in both Kelly and SK-N-BE(2)C cells, respectively (Fig. 5e). The binding of ALYREF to MYCN on chromatin might be due to a yet unknown function of ALYREF in regulating long-range interactions involving chromatin looping.
To confirm the potential interaction between the USP3 promoter sequence binding MYCN at the TSS and the DNA sequence containing the ALYREF-binding site at +6800 bp from the TSS, a chromatin conformation capture (3C) assay ( Fig. 5f) was performed using restriction enzyme BamH1 ( Supplementary Fig. 5c). A specific PCR product (PCR amplicon#1) was detected in BamH1 digested SK-N-BE(2)C samples ( Fig. 5g and Supplementary Fig. 5d). Genomic DNA obtained from SK-N-BE(2)C cells without cross-linking ( Fig. 5f) was used as a positive control (PCR amplicon#2) in these experiments ( Fig. 5f-g and Supplementary Fig. 5d). The DNA looping detected by the 3C assay was further validated using an additional restriction enzyme HindIII, with a restriction site in the ALYREF peak ( Supplementary  Fig. 5c). Restriction digestion with HindIII will prevent the ligation of the MYCN binding and ALYREF-binding sequences, resulting in no PCR product using PCR amplicon #1 primers. As expected, when both HindIII and BamH1 were used, using the same PCR amplicon #1 primers pairs, no PCR product was obtained ( Fig. 5g and Supplementary Fig. 5d). To confirm the 3C data and validate the assay, the PCR products were run on a DNA High Sensitivity (HS) chip using the Agilent 2100 Bioanalyzer (Supplementary Fig 5e). The amplified product was not detected from the sample in which both restriction enzymes used but was present with BamH1 alone (Supplementary Fig 5e). These results suggest that the two DNA fragments are potentially interacting or are in close proximity with each other. To further examine ALYREF as a transcriptional coactivator for MYCN, we evaluated the effect of ALYREF knockdown on USP3 gene transcription in SHEPMYCN3 neuroblastoma cells, which express MYCN under the control of doxycycline, therefore minimising ALYREF knockdown effects on MYCN stability. SHEPMYCN3 neuroblastoma cells were transfected with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2 for 72 h, with and without doxycycline. Transfection of ALYREF siRNA in MYCN-low and ALYREF-low expressing conditions did not affect USP3 mRNA ( Fig. 5h and Supplementary Fig. 5f) or protein (lanes 2-3; Fig. 5i) expression, however, the MYCN-induced increase in USP3 mRNA ( Fig. 5h and Supplementary Fig. 5f) or protein (lanes 5-6; Fig. 5i) levels were diminished by ALYREF knockdown. These data showed that the MYCN-driven transactivation of USP3 was dependent on ALYREF expression. Importantly, we were able to replicate these findings in MYCN-amplified Kelly and SK-N-BE (2)C cells. Transfection with ALYREF siRNA-1 or ALYREF siRNA-2 in SK-N-BE(2)C cells, and ALYREF siRNA-2 in Kelly cells, efficiently knocked down USP3 mRNA (Fig. 5j). Together these data suggest MYCN and ALYREF bind together in a transcriptional coactivator complex to increase USP3 transcript levels, consequently reducing MYCN ubiquitination, and further increasing MYCN protein to the levels required to drive neuroblastoma tumorigenesis.
USP3 regulates MYCN expression through its deubiquitinase activity. Consistent with an MYCN-ALYREF-USP3 signal driving neuroblastoma tumorigenesis, we found that high USP3 mRNA expression in the SEQC neuroblastoma patient cohort (n = 498) also correlated with poor prognosis (Fig. 6a and Supplementary Fig. 6a). Importantly, high USP3 mRNA expression correlated with MYCN expression (Fig. 6b). Multivariate analysis for both overall (Fig. 6c) and event-free ( Supplementary Fig. 6b) survival showed that high USP3 expression had no independent prognostic significance when compared with other prognostic factors such as patient age, clinical stage, and MYCN-amplification. Further analysis for both overall (Fig. 6c) and event-free ( Supplementary Fig. 6c) survival showed that MYCN-amplification and ALYREF expression had the greatest negative effect on USP3's prognostic significance, suggesting that the role of USP3 in the malignant neuroblastoma cell phenotype is dependent on the MYCN-ALYREF regulatory complex. By further examining pan-cancer cell line gene expression (RNA-seq) profiles using Cancer Cell Line Encyclopedia (CCLE)/Cancer Dependency Map 33 we observed that neuroblastoma cell lines had the highest USP3 gene expression compared with all other cancer types ( Supplementary Fig. 6d). We saw significant positive correlations between USP3 and MYCN expression levels ( Supplementary  Fig. 6e) in neuroblastoma tissues. We also observed that neuroblastoma cell lines showed marked dependency on USP3 for cell viability (Fig. 6d), especially in the MYCN-amplified context (Fig. 6e). High endogenous USP3 protein expression was detected in total cellular protein from six MYCN-amplified (lanes 1-6), one MYCN overexpressing (lane 9), and two cMYC expressing cell lines (lanes 7 and 8; Fig. 6f), but not in three neuroblastoma cell lines without MYCN or cMYC expression or normal fibroblasts (lanes 10-13; Fig. 6f). USP3 has been shown to target   3 and 6; Fig. 6g), but not with those using the control IgG antibody (lanes 2 and 4; Fig. 6g). To understand the functional consequence of this interaction, we measured the effects of USP3 expression on MYCN protein levels. We found that MYCN protein levels increased after USP3 overexpression (Supplementary Fig. 6f) in a dose-dependent manner. Furthermore, an ectopically expressed USP3 mutant (Fig. 6h) lacking the catalytic activity (USP3 C168S ) 46,47,49,50 had a significantly reduced effect on MYCN protein expression (Fig. 6i), suggesting that USP3 regulates MYCN expression through its deubiquitinase activity. Although both ectopically expressed USP3 wt and USP3 C168S coimmunoprecipitated with MYCN (Fig. 6j), the half-life of MYCN was only significantly extended in the presence of USP3 wt (lanes 5-8) but not USP3 C168S (lanes 9-12) or an empty (lanes 1-4) expression vector ( Fig. 6k and Supplementary Fig. 6g). Despite their interaction, the USP3 C168S mutant was unable to catalyze the de-ubiquitination of MYCN (lane 5) in HEK293T cells (Fig. 6l), suggesting that the catalytic activity is required for USP3 to impact MYCN ubiquitination levels. Collectively, these results demonstrate that MYCN is a target of USP3.
USP3 is responsible for ALYREF-induced MYCN stability. We further investigated the phenotypic effect of USP3 knockdown in MYCN-amplified neuroblastoma cell lines with high USP3 expression. Silencing endogenous USP3 with USP3 siRNA-1 or or USP3 were transfected with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2 ( Fig. 7f, g). Alamar blue assays for cell viability showed that knocking down ALYREF reduced the number of viable Kelly and SK-N-BE(2)C cells, however, stable overexpression of USP3 reversed the effect of ALYREF knockdown (Fig. 7f). We next examined the hypothesis that the ALYREF effects on MYCN ubiquitination were USP3 dependent. As expected, knocking down ALYREF (lanes 2 and 3; Fig Fig. 8a). Knockdown of ALYREF had profound effects on the cancer phenotype in neuroblastoma cells. Alamar blue assays showed that doxycycline-treated shA-LYREF Kelly and SK-N-BE(2)C cells displayed markedly reduced viability (Fig. 8b) even at lower doxycycline doses (500 ng/ml for SK-N-BE(2)C and 250 ng/ml for Kelly; Supplementary Fig. 8a). In addition, knocking down ALYREF decreased the percentage of Kelly and SK-N-BE(2)C cells in the S phase ( Supplementary  Fig. 8b) and their colony-forming capacity in vitro ( Fig. 8c and Supplementary Fig. 8c). Importantly, these findings were replicated in vivo. SK-N-BE(2)C cells expressing ALYREF shRNA were xenografted into nude mice. Once tumors reached 5 mm in diameter, the mice were divided into doxycycline and vehicle control (Vehicle) subgroups, and fed with 10% sucrose water with or without 2 mg/ml doxycycline 35 , until the tumor reached 1000 mm 3 . Subcutaneous tumor xenografts of SK-N-BE(2)C cells, transduced with doxycycline-inducible ALYREF shRNA, showed significant delays in tumor growth compared to vehicle controls (Fig. 8d). Kaplan-Meier survival analysis showed that doxycycline treatment resulted in longer median survival for mice xenografted with doxycycline-inducible ALYREF shRNA SK-N-BE(2)C cells and treated with doxycycline ( Fig. 8e-f). Immunoblot analysis of snap-frozen tumor tissues showed a decrease in ALYREF, USP3, and MYCN protein expression as a result of doxycycline treatment, compared to vehicle control ( Fig. 8f-g and Supplementary Fig. 8d). These findings demonstrate that ALYREF promotes MYCN-directed tumorigenesis in vitro and in vivo.

Discussion
Transcriptional regulators undergo rapid and timely degradation by ubiquitination in most normal eukaryotic cells. In cancer cells, MYC and MYCN exploit the ubiquitination pathway to stabilize aberrant oncogenic signaling and drive a wide variety of cellular processes required for carcinogenesis. Here, we have described a mechanism for MYCN degradation in MYCN-dependent neuroblastoma, regulated by the ALYREF-USP3 signal. ALYREF activates gene transcription by enhancing DNAbinding and activity [16][17][18] of transcription factors on target gene promoters. Our data suggest that the interaction between ALYREF and MYCN is likely to be important for efficient gene transcription. We confirmed that ALYREF protein directly binds to MYCN protein at the small region of amino acids 281-345  5 ALYREF and MYCN form a transcriptional activator complex which upregulates USP3 expression. a Approach to identifying ALYREF target gene candidates in the context of MYCN stability regulation. ALYREF ChIP-seq data was used to analyze 3598 ALYREF-bound chromatin regions. b Genomewide profile of ALYREF-chromatin interactions. c Integrative Genomic Viewer (IGV) representation of the USP3 gene promoter detailing MYCN and ALYREF-binding motifs. d Schematic representation of the USP3 and negative control target sequences used for ChIP-PCR, detailing MYCN and ALYREFbinding motifs and their distance from the transcription start site (TSS). e Single ChIP assays for MYCN or double ChIP assays for MYCN:ALYREF complex analyzed by RT-PCR (n = 3 per group) with primers targeting the negative control (Control) region or the USP3 gene. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in expression levels were compared to the negative control (Control). f Schematic of chromatin conformation capture (3C) assay used to detect the interactions between the DNA fragments containing the MYCN binding site at the USP3 promoter region (red, 0 bp) and that containing the ALYREF peak (green, +6800 bp from USP3 TSS). g Representative agarose gel from 3C assay performed with BamH1 and/or HindIII showing PCR product of MYCN and ALYREF interaction (PCR Amplicon #1; 166 bp) and the positive control non-cross-linked DNA (PCR Amplicon #2; 53 bp). h RT-PCR (n = 3 per group) and i immunoblot analyses of USP3 (h) and USP3, ALYREF, and MYCN (i) expressions following siRNA-mediated ALYREF knockdown in SHEPMYCN3 cells. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in expression levels were compared to the vehicle-treated control siRNA. j RT-PCR analysis (n = 3 per group) of USP3 expression following siRNA-mediated ALYREF knockdown. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in expression levels were compared to the control siRNA.   56 . We conclude that the observed ALYREF binding is biologically significant in the context of malignant cells, however, the exact role of ALYREF in regulating MYCN-dependent or -independent gene transcription in neuroblastoma remains largely undetermined. Direct pharmacological approaches to inhibition of MYCN have been notoriously difficult. Alternatively, MYCN upstream positive regulators, especially enzymes, may be better targets for treatment. Indeed, it has been demonstrated that inhibitors of E3 ubiquitin ligases AURKA 59-61 and PLK1 62 promote MYCN protein degradation. Deubiquitinating enzymes (DUBs) are increasingly recognized as attractive targets for therapeutic intervention as inhibition of their activity is a more direct approach than activating E3 ligases. To date, only two DUBs have been implicated in negative regulation of MYCN, USP28 57 , and USP7 43 . The USP7 specific DUB-inhibitor P22077 already has been introduced as a potential therapeutic agent for MYCN-driven neuroblastoma 43 . Destabilizing MYCN represents an emerging therapeutic strategy, however, there is still a need to identify novel molecular mechanisms allowing the design of inhibitors that more potently and selectively reduce MYCN protein levels and activity. In our study, we identified USP3 as an MYCN DUB, further expanding our knowledge of the diversity of MYCN protein homeostasis. Here we show that USP3 cleaves K-63linked polyubiquitin chains off MYCN and can regulate MYCN protein turnover. Although the K-48 linkage on its own acts as a mark for proteasomal degradation of MYCN, the function of K-63 chains in MYCN protein degradation is less clear 40 . We argue that downregulation of USP3 through genetic inhibition of the ALYREF-MYCN protein complex disturbed MYCN protein homeostasis that also changed the abundance of K-48-linked polyubiquitin chains on MYCN and led to proteasomal degradation. Interestingly, previous studies have shown that K63ubiquitination of MYC by Huwe1 63 , HectH9 64 , or Skp2 65 stimulates the transcriptional activity of MYC in addition to regulating its turnover.
ALYREF expression is dysregulated in several cancer types 22,23 . Studies have implicated ALYREF as a regulator of cancer cell growth, since depletion of ALYREF results in a significant decrease in cell proliferation 24 . Furthermore, ALYREF knockdown caused a significant decrease in cancer cell migration 23 . Our in vitro and in vivo data strongly supports the role of ALYREF as a co-driver for oncogenesis in neuroblastoma. We also suggest that the regulatory relationship described in this work between ALYREF and MYCN may extend to other human cancer types with high MYC/MYCN dependency (medulloblastoma, retinoblastoma, hematologic malignancies, glioblastoma, ovarian cancer, Wilms' tumor, neuroendocrine prostate cancer). High ALYREF expression in neuroblastoma tumor tissue had prognostic significance, which was independent of MYCN-amplification, therefore we argue that ALYREF may have MYCNindependent oncogenic effects. Despite considerable analysis carried out in this study on the ALYREF and MYCN regulatory circuit, it is not yet fully clear how MYCN-promoted oncogenic levels of ALYREF reprogram cells to help the cancer state.
Our study extends the knowledge regarding genes crucial for tumor progression in 17q21-ter gain neuroblastoma patients, and indicates that the ALYREF-USP3-MYCN regulatory complex is valuable as a potential target for the treatment of neuroblastoma by exploiting the enforced ALYREF addiction present in MYCNamplified cells.   6 USP3 expression associates with MYCN-amplification and predicts poor patient survival. a Overall survival of patients in the SEQC cohort (n = 498) when dichotomized by median USP3 expression (RNA-Seq). Hazard ratios (HR) and two-sided log-rank p-values are presented from a univariate CoxPH model. b Scatter plot, with a linear regression fit for the SEQC cohort (n = 498) for MYCN vs USP3 expression (RNA-Seq, log 2 RPM). A two-sided Pearson's correlation test was used to derive p-values. c Multivariate overall survival (n = 498) dichotomized for advanced stage of disease, age of patient, MYCN amplification, ALYREF and USP3 expression using cox regression modeling, p-values from two-sided log-rank tests. d USP3 CRISPRi dependency expressed as CERES scores across 27 cancer types, p-value is from a two-sample t-test between neuroblastoma cell lines (n = 718) and an aggregate of all other cell lines. e USP3 CERES scores for neuroblastoma (NB) divided into MYCN-non-amplified (n = 4) and MYCN-amplified (n = 15) cell lines, p-value is from a two-sample t-test. f Immunoblot for MYCN and USP3 in a range of neuroblastoma cells. g Immunoblot for USP3 after immunoprecipitation of MYCN. h Schematic of wild-type (USP3 wt ) and mutant (USP3 C168S ) USP3 proteins. ZnF zinc-finger ubiquitin-binding domain, USP ubiquitin-specific protease domain. i Immunoblot for USP3 wt andUSP3 C168S proteins and HA-MYCN from HEK293T cells. j Immunoblot for Flag-USP3 wt or Flag-USP3 C168S after immunoprecipitation of HA-MYCN. Vector indicates empty vector control. k HEK293T cells expressing USP3 wt and USP3 C168S were treated with cycloheximide (CHX) and subjected to immunoblot for MYCN and Flag-USP3. l HEK293T cells expressing HA-MYCN, His-Ubiquitin, USP3 wt , and USP3 C168S or vector control were subjected to HA-MYCN immunoprecipitation and immunoblot for ubiquitination. The dot (a measure of center for the error bars) in each row represents the mean hazard ratio, while the error bars represent the 95% confidence intervals for the presented mean hazard ratio in c. Neuroblastoma cell line NBLS was kindly provided by Prof. Susan L. Cohn (Northwestern University, Chicago, IL, USA). Neuroblastoma cell line LAN-1 was kindly provided by Dr. John Maris (Children's Hospital of Philadelphia, Philadelphia, USA). MRC5 and WI-38 normal human fibroblasts were purchased from ATCC. All cell lines used were authenticated by Cell Bank Australia (Westmead, NSW, Australia), free from mycoplasma, and cultured at 37°C/5% CO 2 in a humidifier incubator. All cells were freshly thawed from initial seeds and were not cultured for more than 2 months.

Experimental models and subject details
Establishment of doxycycline-inducible control shRNA and ALYREF shRNAexpressing neuroblastoma cell lines. To generate control and stable ALYREFknockdown cells, SK-N-BE(2)C and Kelly cells were infected with pTRIPZ non-targeting control vector (RHS5087-EG10189) or pTRIPZ encoding an ALYREF-specific shRNA (5′-AGAGTTCCTGAATATCGGC-3′) purchased from Dharmacon (Millenium Science Australia, VIC, Australia). Lentiviral vectors were packaged in HEK293 cells according to the manufacturer's protocol. 48 h later, the viral supernatant was incubated with polybrene (Santa Cruz All experimental work was conducted in biosafety or cytotoxic cabinets to further protect the animals from the microbiological threat. Female Balb/c nude mice aged 5-6 weeks were injected subcutaneously with 2 × 10 6 doxycycline-inducible ALYREF shRNA BE(2)C cells into the right flank. When engrafted tumors reached 5 mm size in diameter, tumor-bearing mice were randomized into two groups: vehicle (DMSO)-treated and doxycycline (DOX)-treated mice. Mice with undetectable engraftment were excluded. Mice were fed with 5% sucrose water with or with doxycycline (Sigma) at 2 mg/mL. Tumor development was monitored and tumor volume was calculated using (length × width × height)/2. Mice were culled when tumor volume reached 1 cm 3 , and tumor tissues were snap-frozen and analyzed by immunoblotting for ALYREF, MYCN, USP3, and actin protein expression.
Microarray analysis on TH-MYCN +/+ ganglia tissues. Ganglia from TH-MYCN +/+ mice and wild-type controls were collected at 2, 4, and 6 weeks of age. RNA was extracted from the cells with PureLink RNA Kit (Life Technologies), and genomewide differential gene expression was examined using Agilent SurePrint G3 Mouse GE 8×60K Microarrays. Patient tumor genome/gene expression/survival analyses. Whole-genome sequencing (WGS) data, containing normalized coverage and ploidy calls for genomic segments in 135 clinically annotated primary neuroblastoma tumors (TARGET neuroblastoma cohort), were obtained through the TARGET data matrix (https:// ocg.cancer.gov/programs/target/data-matrix) and further processed using the R statistical language and R Studio. Average ploidy values were calculated in 10 kb bins across chromosome 17 using the Genomic Ranges R package (v1.40.0), wherein an average ploidy (n), n ≤ 1 was considered a loss, 1 ≤ n ≤ 3 was considered diploid and n ≥ 3 considered to be a gain 25 . Average ploidy across the 17q21-ter region was then calculated in each sample and classified according to the above strata. We utilized matched clinical annotations containing event-free and overall survival data to construct Kaplan-Meier survival curves of the different molecular subgroups using the survminer (v0.4.6) and survival R packages (v2.42.1). RNA-seq data, in the form of gene-level counts, for 90 samples, which had paired WGS data (aforementioned), were also obtained from the TARGET data matrix. We performed differential gene expression testing between 17q21-ter gain vs diploid as well as MYCN-amplified vs non-amplified samples using the DESeq2 R package (v1.22.0) 66 and filtered for genes passing an adjusted p-value threshold of <0.05 and log2 fold-change threshold being ≥ the lower quartile of all positive log2foldchange values passing the adjusted p-value threshold. Correlation of gene expression between genes of interest as well as with gene copy number was analyzed using Pearson correlation and resultant p-values were adjusted using the Bonferroni method, followed by filtering using an adjusted p-value threshold of <0.05 and r > 0. Using the survival R package (v2.42.1) univariate Cox proportional hazards (CoxPH) regression models for genes of interest were performed and then filtered based on a hazard ratio (HR) threshold of >1 and individual log-rank test p-value < 0.05. RNA-seq data, expressed in reads per million (RPM) for each gene, for 498 clinically annotated primary neuroblastoma samples (SEQC neuroblastoma cohort) were obtained from the gene expression omnibus (GEO) with the accession GSE62564 31 . Microarray data, for each gene, for 476 neuroblastoma samples (Kocak neuroblastoma cohort), were downloaded from R2 platform (http://r2.amc. nl). We utilized matched clinical annotations containing event-free and overall survival data to construct Kaplan-Meier survival curves of subgroups dichotomized by median gene expression using the survival R package (v2.42.1). Using the same R package (v2.42.1), we ran either univariate or multivariate CoxPH regression models for genes of interest, regressing covariates associated with neuroblastoma prognosis such as advanced disease stage (3 and 4), age at diagnosis (>18 months), and MYCN-amplification status. All statistical tests concerning Kaplan-Meier analyses were done utilizing log-rank tests, adjusted using the Bonferroni method for multiple hypotheses testing where appropriate.
RNA isolation and quantitative real-time PCR. Total RNA was extracted using the PureLink RNA kit (Life Technologies) according to the manufacturer's protocol. 1 μg total RNA was reverse-transcribed using the Tetro cDNA synthesis kit (Life Technologies). Gene expression was verified using Power SYBR Green mix (Applied Biosystems, Life Technologies) performed on ABI7500 thermo-cycler (Applied Biosystems, Life Technologies) with a standard protocol. Differential gene expression was measured using the log2ΔΔCt analysis. All mRNA expression levels were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) or Beta- Fig. 7 USP3 is responsible for ALYREF-driven MYCN ubiquitination. a Immunoblot analysis of MYCN and USP3 expression in Kelly and SK-N-BE(2)C cells following siRNA-mediated USP3 knockdown for 48 h. b Cell viability (n = 3 per group) (c) and proliferation (n = 3 per group) analysis of Kelly and SK-N-BE(2)C cells transfected with USP3 siRNA-1, USP3 siRNA-2, or Control siRNA for 96 h. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in cell growth were compared to Control siRNA expressing cells. d Kelly and SK-N-BE(2)C cells transfected with USP3 siRNA-1, USP3 siRNA-2, or Control siRNA were grown for 12 days (Kelly) or 10 days (SK-N-BE(2)C), followed by colony formation assay. e Immunoblot analysis of USP3 expression in stable Kelly and SK-N-BE(2)C cells overexpressing USP3 (USP3) or Vector control (Vector). NTC, non-transduced control cells. f Stable Kelly and SK-N-BE(2)C cells overexpressing USP3 (USP3) or Vector control (Vector) were co-transfected with ALYREF siRNA-1 or ALYREF siRNA-2 or Control siRNA, followed by cell viability measurements (n = 3 per group) at 72 h. Two-sided unpaired Student's t-tests were performed to derive pvalues. Differences in cell viability were compared to Vector Control siRNA expressing cells. g Stable Kelly and SK-N-BE(2)C cells overexpressing USP3 (USP3) or Vector control (Vector) were co-transfected with ALYREF siRNA-1, or ALYREF siRNA-2 or control siRNA, followed by treatment with MG132 for 4 h. Cells were then subjected to endogenous MYCN immunoprecipitation and immunoblot analyses for K-63 and K-48-linked ubiquitination. Immunoprecipitation. For endogenous MYCN or ALYREF immunoprecipitation assays, cells were lysed in cold NP40 buffer (Sigma) supplemented with protease inhibitor (Sigma). For endogenous USP3 immunoprecipitation assays, cells were lysed in cold BC100 buffer (Sigma) supplemented with protease inhibitor (Sigma). 5% of the cell extract was saved as the input, and the rest (750 μg total protein) was incubated with either 10 μg MYCN-specific antibody (Merck Millipore, VIC, Australia), 10 μg ALYREF-specific antibody (D3R4R; Cell Signaling Technologies, Beverly, MA, USA), or with 10 μg control IgG antibody (mouse from Santa Cruz Biotechnology and rabbit from Cell Signaling) and A/G PLUS agarose beads (Santa Cruz Biotechnology) at 4°C overnight. After three washes with the lysis buffer, the bound proteins were eluted by boiling with SDS sample buffer. Bound proteins were resolved by SDS-PAGE.
For immunoprecipitation of ectopically expressed Flag and HA-tagged proteins, HEK293T human embryonic cells were transiently co-transfected with an HA-MYCN-expression pcDNA3.1 construct, together with an empty vector or   Flag-ALYREF-expression pCMV6 construct or Flag-USP3-expression pCMV3Tag3A construct for 24 h. Cells were lysed in cold NP40 or BC100 buffer supplemented with protease inhibitor (Sigma), respectively. 5% of the cell extract was saved as the input, and the rest (500 μg total protein) was incubated with either 15 μg of Flag (DYKDDDDK)-specific antibody (9A3; Cell Signaling Technologies), 15 μg HAspecific antibody (3F10; Sigma) or with 15 μg control IgG (Santa Cruz Biotechnology) and A/G PLUS agarose beads (Santa Cruz Biotechnology) at 4°C overnight. After three washes with the lysis buffer, the bound proteins were eluted by boiling with SDS sample buffer. For HA-MYCN deletion mutant experiments, pcDNA3.1-HA-MYCN plasmid or the mutant constructs were co-transfected with the construct expressing ALYREF for 24 h, and the cells were lysed in NP40 buffer (Sigma). 5% of the cell extract was saved as the input, and the rest (500 μg total protein) were incubated with either 15 μg of Flag (DYKDDDDK)-specific antibody (9A3; Cell Signaling Technologies) or with 15 μg control IgG (Santa Cruz Biotechnology) and A/G PLUS agarose beads (Santa Cruz Biotechnology) at 4°C overnight. After three washes with the lysis buffer, the bound proteins were eluted by boiling with SDS sample buffer. Bound proteins were resolved by SDS-PAGE.
Isolation of nuclear and cytoplasmic extract. The nuclear and cytoplasmic extracts were prepared using a NE-PER Nuclear Cytoplasmic Extraction Reagent kit (ThermoFisher Scientific) according to the manufacturer's instruction. Briefly, cells were washed twice with cold PBS and centrifuged at 500×g for 3 min. The cell pellet was suspended in 200 μl of cytoplasmic extraction reagent I by vortexing. The suspension was incubated on ice for 10 min followed by the addition of 11 μl of a second cytoplasmic extraction reagent II, vortexed for 5 s, incubated on ice for 1 min, and centrifuged for 5 min at 16,000×g. The supernatant fraction (cytoplasmic extract) was transferred to a pre-chilled tube. The insoluble pellet fraction, which contains crude nuclei, was resuspended in 100 μl of nuclear extraction reagent by vortexing for 15 s and incubated on ice for 10 min, then centrifuged for 10 min at 16,000×g. The resulting supernatant, constituting the nuclear extract, was used for the subsequent experiments along with the cytoplasmic extract.
MYCN stability analysis. SK-N-BE(2)C cells were transfected with pCMV6-ALYREF plasmid DNA for 48 h or stable ALYREF knockdown SK-N-BE(2)C and Kelly cells were induced with doxycycline for 72 h, then treated with 100 μg/ml cycloheximide (Sigma) for 0-60 min. Cells were collected at 15 min treatment intervals and protein lysates were analyzed by Western blot to measure changes in MYCN protein stability.
HEK293T cells were co-transfected with pCMV3Tag3A-Flag-USP3 WT or pCMV3Tag3A-Flag-USP3 C168S mutant constructs along with pcDNA3.1-HA-MYCN plasmid DNA for 24 h, then treated with 100 μg/ml cycloheximide (Sigma) for 0-60 min. Cells were collected at 15 min treatment intervals and protein lysates were analyzed by western blot to measure changes in MYCN protein stability.
Ubiquitin assays. Cells were exposed to the proteasomal inhibitor MG132 (Sigma) at 20 μM concentration 42 . Four hours post MG132 treatment, cells were lysed with denaturing lysis buffer (6 M guanidine-HCl, 50 mM sodium phosphate, pH 8.0, 500 mM NaCl, 5 mM imidazole). 5% of the protein was saved to serve as an input. The rest of the sample (750 μg total protein) was diluted in BC100 buffer (Sigma) freshly supplemented with protease inhibitor cocktail (Sigma) and subjected to immunoprecipitation with either 5 μg MYCN-specific antibody (Merck Millipore), or with 5 μg control mouse IgG antibody (Santa Cruz Biotechnology) and A/G PLUS agarose beads (Santa Cruz Biotechnology) at 4°C overnight. Prior to immunoprecipitation, the beads were washed three times with denaturing wash buffer (8 M urea, 50 mM sodium phosphate, pH 8.0, 500 mM NaCl). After overnight incubation at 4°C, beads were washed five times with cold BC100 buffer (Sigma) and once with cold sodium chloride solution (150 mM). Proteins were eluted in Laemmli sample buffer (62.5 mM Tris-HCl, pH 6.8; 25% glycerol; 2% SDS; 0.01% Bromophenol Blue) supplemented with β-mercaptoethanol (Bio-Rad) at a final concentration of 5%, and separated by electrophoresis using 10.5% or 10-14% Tris-HCl Criterion gels (Bio-Rad).
Chromatin-immunoprecipitation (ChIP) assays and double-antibody ChIP assays. Preparation of DNA from SK-N-BE(2)C and Kelly cells for ChIP assay was performed using the Chromatin-Immunoprecipitation Assay kit (Merck Millipore) according to the manufacturer's instructions. ChIP assays were performed with mouse anti-MYCN antibody (B84B; Santa Cruz Biotechnology); anti-ALYREF antibody (11G5; ImmunoQuest) or control mouse IgG antibody (Santa Cruz Biotechnology). DNA was purified using a MiniElute PCR Purification Kit (Qiagen). Real-time PCR was performed with primers designed to cover the regions of the ALYREF and USP3 genes containing MYC-responsive or ALYREF-responsive motifs or remote negative control regions. Fold enrichment of the ALYREF or USP3 genes containing the binding regions by the anti-MYCN or anti-ALYREF antibody was calculated by dividing the PCR product from this region by the PCR product from the negative control region, relative to the input.
Chromosome conformation capture (3C) assay. For 3C assay 67 , nuclei were extracted from fixed SK-N-BE(2)C cells as described in the chromatinimmunoprecipitation protocol. BamH1 and HindIII (Promega Australia, NSW, Australia) restriction enzymes were selected based on in silico analysis to cut DNA. The digested nuclei were then treated with T4 DNA ligase (ThermoFisher Fig. 8 ALYREF is required for the growth and tumorigenicity of MYCN-amplified neuroblastoma cells. Kelly and SK-N-BE(2)C cells expressing ALYREF shRNA (shALYREF) were treated with doxycycline (DOX; 2 µg/ml) for 72 h, then subjected to immunoblot analysis of MYCN and ALYREF expression (a) and cell viability (b) measurements (n = 3 per group). Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in cell viability were compared to the vehicle control shRNA (vehicle). c Kelly and SK-N-BE(2)C cells expressing ALYREF shRNA (shALYREF) or control shRNA (shControl) were treated with doxycycline (DOX) for 12 days (Kelly) or 10 days (SK-N-BE(2)C), followed by colony formation assay (n = 3 per group). Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in colony formation were compared to the vehicle control shRNA (vehicle). SK-N-BE(2)C cells expressing ALYREF shRNA (shALYREF) were xenografted into nude mice. Once tumors reached 5 mm in diameter, the mice were divided into doxycycline (DOX) and vehicle control (Vehicle) subgroups and fed with 10% sucrose water, with or without 2 mg/ml doxycycline. d Tumor volume was taken from each mouse (n = 8 per treatment group), starting from day 0 post-treatment, until the tumor reached 1000 mm 3 . The effect of DOX on tumor progression was evaluated using two-way ANOVA e Kaplan-Meier survival curves showed the probability of overall survival of the mice (n = 8). A two-sided long-rank test was used to obtain p-value. Immunoblot (f) and protein densitometry (g) analysis of tumor samples (n = 8) from SK-N-BE(2)C tumor xenografts, analyzed for ALYREF, USP3, and MYCN expression. Phox2B expression was used as a neuroblastoma marker and loading control. Two-sided unpaired Student's t-tests were performed to derive p-values. Differences in ALYREF and MYCN protein expressions were compared to the vehicle-treated ALYREF shRNA subgroup. Respective p-values (p) are displayed. Comparisons were not significant unless otherwise noted. Data is representative of three independent experiments with similar results in a and c. Data are shown as mean ± s.e.m. (error bars) and representative of three independent experiments in b and c and 8 mice per treatment group in d. In vivo data are representative of two experiments in f and quantification is shown as mean ± s.e.m. (error bars) of each treatment group in g. Scientific) to ligate DNA fragments that interacted with each other. DNA fragments after ligation were purified using phenol/chloroform/isoamyl alcohol. Then PCR was performed for 35 cycles using the following PCR conditions: 94°C for 10 min; and thermocycling at 94°C for 30 s, 60°C for 30 s and 72°C for 30 s; following by 72°C 15 min. PCR product was loaded on a 1.5% agarose gel with SYBR Safe (Life Technologies), and images were taken using a ChemiDoc Imaging System (Bio-Rad). PCR samples from the 3C assay were also run on a DNA High Sensitivity (HS) chip using the Agilent 2100 Bioanalyzer (Agilent Technologies Australia, VIC, Australia) according to the manufacturer's instructions. Uncropped PCR gel scans are provided as Supplementary Information.
Cell proliferation assay. Cell proliferation was measured using the BrdU ELISA kit (Roche, Australia) according to the manufacturer's instructions. Cells were grown in a medium containing 5 μg/ml BrdU for 4 h and fixed in 4% paraformaldehyde. DNA was denatured, and cells were permeabilized in 2 N HCl with 0.5% Triton X-100 (Sigma) and then blocked with 5% BSA in PBS. Anti-BrdU (1:200) was added following the manufacturer's protocol. After washing with 5% BSA in PBS, the cells were incubated with Alexa-Fluor-594-conjugated anti-mouse-IgG (1:100). Changes in cell proliferation were calculated from the absorbance readings at 370 nm (490 nm reference wavelength) on the Benchmark Plus microplate reader (Bio-Rad).
Colony formation. 200 SHEPMYCN3 cells transfected with control siRNA, ALYREF siRNA-1, or ALYREF siRNA-2 were seeded in 6-cm 2 plates and treated with and without doxycycline (Sigma) for 12 days. 250 SK-N-BE(2)C and 500 Kelly cells expressing control siRNA, USP3 siRNA-1, or USP3 siRNA-2 were seeded in 6cm 2 plates and grown for 10 and 12 days, respectively. 250 SK-N-BE(2)C and 500 Kelly cells expressing shALYREF or control shControl constructs were seeded in 6cm 2 plates and kept under doxycycline treatment for 10 and 12 days, respectively, to allow colonies to form. Colonies were fixed and stained with crystal violet solution (Sigma) then washed with water to remove the unincorporated stain. Cells were photographed and colony formation was quantified from crystal violet absorbance readings at 590 nm on the Benchmark Plus microplate reader (Bio-Rad). Uncropped colony formation assay scans are provided as Supplementary Information.
Cell viability assay. Cell viability was measured using the Alamar Blue assay (Life Technologies) according to the manufacturer's protocol. Signals were quantitated on a VICTOR 2 Multilabel counter (Perkin Elmer Australia) at an excitation wavelength of 560 nm and an emission wavelength of 590 nm.

Data availability
The ALYREF ChIP-seq data have been deposited at the Gene Expression Omnibus Website with series number GSE150303. We also obtained several other publicly available ChIP-seq data sets (GSE80151) of MYCN, RNA Polymerase II, BRD4, H3K27ac, H3K4me3, and H3K27me3 to complement our ALYREF ChIP-seq data (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE80151), fastq files were obtained directly from the European Nucleotide Archive (ENA) under the study accession PRJNA318044. Gene expression and relevant patient prognosis information in TARGET, SEQC, and Kocak neuroblastoma patient data sets were downloaded from R2 platform (http://r2.amc.nl). Whole-genome sequencing (WGS) data were obtained through the TARGET data matrix (https://ocg.cancer.gov/programs/target/data-matrix). RNA-seq data that had paired WGS data, were also obtained from the TARGET data matrix. RNA-seq data for the SEQC neuroblastoma cohort were obtained from the gene expression omnibus (GEO) with the accession GSE62564. We utilized public data resources produced by the Cancer Cell Line Encyclopedia (CCLE) and Project Achilles via the Cancer Dependency Map (DepMap, 20Q1) portal (https://depmap.org/portal/). Data underlying all figures and supplementary figures, uncropped and unprocessed immunoblot scans, as well as colony formation assay and PCR agarose gel pictures for all figures and supplementary figures, are provided as Source Data. All other relevant data are available from the corresponding authors on request.