Therapeutic targeting of YY1/MZF1 axis by MZF1-uPEP inhibits aerobic glycolysis and neuroblastoma progression

As a hallmark of metabolic reprogramming, aerobic glycolysis contributes to tumorigenesis and aggressiveness. However, the mechanisms and therapeutic strategies regulating aerobic glycolysis in neuroblastoma (NB), one of leading causes of cancer-related death in childhood, still remain elusive. Methods: Transcriptional regulators and their downstream glycolytic genes were identified by a comprehensive screening of publicly available datasets. Dual-luciferase, chromatin immunoprecipitation, real-time quantitative RT-PCR, western blot, gene over-expression or silencing, co-immunoprecipitation, mass spectrometry, peptide pull-down assay, sucrose gradient sedimentation, seahorse extracellular flux, MTT colorimetric, soft agar, matrigel invasion, and nude mice assays were undertaken to explore the biological effects and underlying mechanisms of transcriptional regulators in NB cells. Survival analysis was performed by using log-rank test and Cox regression assay. Results: Transcription factor myeloid zinc finger 1 (MZF1) was identified as an independent prognostic factor (hazard ratio=2.330, 95% confidence interval=1.021 to 3.317), and facilitated glycolysis process through increasing expression of hexokinase 2 (HK2) and phosphoglycerate kinase 1 (PGK1). Meanwhile, a 21-amino acid peptide encoded by upstream open reading frame of MZF1, termed as MZF1-uPEP, bound to zinc finger domain of Yin Yang 1 (YY1), resulting in repressed transactivation of YY1 and decreased transcription of MZF1 and downstream genes HK2 and PGK1. Administration of a cell-penetrating MZF1-uPEP or lentivirus over-expressing MZF1-uPEP inhibited the aerobic glycolysis, tumorigenesis and aggressiveness of NB cells. In clinical NB cases, low expression of MZF1-uPEP or high expression of MZF1, YY1, HK2, or PGK1 was associated with poor survival of patients. Conclusions: These results indicate that therapeutic targeting of YY1/MZF1 axis by MZF1-uPEP inhibits aerobic glycolysis and NB progression.


Introduction
Neuroblastoma (NB) is the most common extracranial solid malignancy in pediatric population, and accounts for approximately 15% of all childhood cancer deaths [1]. Despite advances in molecular mechanisms and multimodal therapy [2,3], the clinical course of high-risk NB cases still remains unfavorable, and is featured by rapid progression and high mortality [1]. As a hallmark of metabolic reprogramming, even in the presence of oxygen, tumor cells uptake and convert a large amount of glucose into lactic acid to support their tumorigenecity and aggressiveness, which is known as aerobic glycolysis or Warburg effect [4][5][6][7]. High rates of glycolysis are consistently observed in most of tumors, accompanied by up-regulation of glycolytic enzymes such as hexokinase 2 (HK2), phosphoglycerate kinase 1 (PGK1), and enolase 1 (ENO1) [8,9]. Meanwhile, small organic molecules, such as 3-bromopyruvate or 2-deoxyglucose (2-DG), are able to inhibit aerobic glycolysis and exhibit therapeutic potential for repressing tumor progression [10,11]. Thus, it is important to investigate the mechanisms and therapeutic strategies for aerobic glycolysis during tumor progression.
Recent studies show that aerobic glycolysis is driven by activation of oncogenes or inactivation of tumor suppressors. For example, hypoxia inducible factor 1 alpha, a key mediator of hypoxic response, contributes to aerobic glycolysis by up-regulating glucose transporters 1 (GLUT1) and lactate dehydrogenase A (LDHA) [12]. Onocgenic c-Myc facilitates glycolysis process through inducing HK2 and LDHA expression [13,14]. Meanwhile, p53 represses aerobic glycolysis through reducing promoter activity of GLUT1 and GLUT4 [15]. Long noncoding RNA LINC01554 inhibits aerobic glycolysis via promoting degradation of pyruvate kinase isozyme M2 (PKM2) in hepatocellular carcinoma cells [7]. However, the mechanisms regulating the expression of glycolytic genes in NB still remain to be determined.
In this study, through an integrative screening approach, we identify myeloid zinc finger 1 (MZF1) and its upstream open reading frame (uORF)-derived peptide (uPEP) as crucial regulators of aerobic glycolysis and NB progression. We demonstrate that MZF1 is up-regulated in NB tissues and cells, and facilitates the aerobic glycolysis, growth, and aggressiveness of NB cells by up-regulating HK2 and PGK1. Meanwhile, MZF1-uPEP interacts with Yin Yang 1 (YY1) to repress its transactivation, resulting in transcriptional inhibition of MZF1 and downstream glycolytic genes. Pre-clinically, administration of a cell-penetrating MZF1-uPEP or lentivirus over-expressing MZF1-uPEP significantly suppresses aerobic glycolysis, tumorigenesis and aggressiveness, indicating the crucial roles of MZF1-uPEP in repressing YY1/MZF1 axis during NB progression.

Real-time quantitative RT-PCR (qRT-PCR)
Total RNA was isolated with RNeasy Mini Kit (Qiagen Inc., Valencia, CA). Reverse transcription reactions were conducted with Transcriptor First Strand cDNA Synthesis Kit (Roche, Indianapolis, IN). Real-time PCR was performed with SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and primers (Table S1).

Rescue of target gene expression
To restore target gene expression induced by MZF1 or MZF1-uORF, tumor cells were transfected with shRNAs targeting HK2 and PGK1, or YY1 expression vector. To rescue gene expression altered by knockdown of MZF1 or MZF1-uORF, HK2 and PGK1 expression vectors or shRNAs specific for YY1 (Table S3) were transfected into tumor cells with  Lipofectamine 3000 (Life Technologies, Inc., Gaithersburg, MD).

Design and synthesis of cell-penetrating peptides
Cell-penetrating peptide of MZF1-uPEP was designed and synthesized (ChinaPeptides Co. Ltd, Shanghai, China). The 11-amino acid long peptide (YGRKKRRQRRR) from Tat protein transduction domain served as a cell-penetrating peptide. Thus, inhibitory peptides were chemically synthesized by linking with biotin-labeled cell-penetrating peptide at N-terminus and conjugating with fluorescein isothiocyanate (FITC) at C-terminus, with purity larger than 95%.

Biotin-labeled peptide pull-down assay
Cellular proteins were isolated using 1× cell lysis buffer (Promega), and incubated with biotin-labeled peptide at 4°C overnight. Then, incubation of cell lysates with streptavidin-agarose was undertaken at 4°C for 2 hrs. Beads were extensively washed, and proteins pulled down were measured by western blot assay.

Sucrose gradient sedimentation
Tumor cells were treated with 100 μg/ml of cycloheximide (Sigma) for 5-10 min. Cell extracts were layered on top of 15-30% (w/v) linear sucrose gradient. After centrifugation at 40,000 ×g for 2 hrs at 4°C, fractions were collected using a piston-gradient fractionator (Biocomp, Fredericton, Canada). The polysome profiles were monitored by absorbance of light with a wavelength of 260 nm (A260). The polysome-bound transcripts were extracted and detected by real-time qRT-PCR.

In vivo tumorigenesis and aggressiveness assays
All animal experiments were carried out in accordance with NIH Guidelines for the Care and Use of Laboratory Animals, and approved by the Animal Care Committee of Tongji Medical College (approval number: Y20080290). In vivo tumor growth (1×10 6 tumor cells per mouse) and experimental metastasis (0.4×10 6 tumor cells per mouse) studies were performed with blindly randomized four-week-old female BALB/c nude mice as previously described [16][17][18][19][20]. For in vivo therapeutic studies, tumor cells (1×10 6 or 0.4×10 6 ) were injected into dorsal flanks or tail vein of nude mice, respectively. One week later, mice were blindly randomized and treated by tail vein injection of synthesized cell-penetrating peptide (ChinaPeptides, Shanghai, China) or lentivirusmediated MZF1-uORF (1×10 7 plaque-forming units in 100 μl phosphate buffer saline). Animals were imaged using the In-Vivo Xtreme II small animal imaging system (Bruker Corporation, Billerica, MA).

Patient tissue samples
The Institutional Review Board of Tongji Medical College approved the human tissue study (approval number: 2011-S085). All procedures were undertaken in accordance with guidelines set forth by Declaration of Helsinki. Written informed consent was obtained from all legal guardians of patients. Patients with a history of preoperative chemotherapy or radiotherapy were excluded. Human normal dorsal root ganglia tissues were collected from therapeutic abortion. Fresh specimens were collected at surgery, validated by pathological diagnosis, and stored at -80°C.

Statistical analysis
All data were shown as mean ± standard error of the mean (SEM). Cutoff values were determined by average gene expression levels. Student's t test, one-way analysis of variance (ANOVA), and χ 2 analysis were applied to compare the difference in tumor cells or tissues. Fisher's exact test was applied to analyze the statistical significance of overlap between two gene lists. Pearson's correlation coefficient was applied for analyzing the relationship among gene expression. Log-rank test and Cox regression analysis were used to assess survival difference and hazard ratio. All statistical tests were two-sided and considered statistically significant when false discovery rate-corrected P values less than 0.05.

MZF1 facilitates the transcription of glycolytic genes in NB
To investigate transcriptional regulators of glycolytic gene expression and tumor progression, we performed comprehensive analysis of a public dataset of 88 NB cases (GSE16476) [24], and identified 9, 9, and 7 glycolytic genes (P<0.05) differentially expressed in NB specimens with varied status of age, death, and international neuroblastoma staging system (INSS) stages, respectively ( Figure 1A). Based on over-lapping analysis of these results (P<0.001), 7 genes were found to be consistently associated with age, death, and advanced INSS stage of NB ( Figure 1A and Table S4). Similarly, we also found 9 transcription factors consistently associated with these clinical features of 88 NB cases ( Figure 1A and Table S4), which were subjective to further over-lapping analysis with potential transcription factors regulating all of 7 genes revealed by Genomatix program (http://www.genomatix.de). The results revealed that two transcription factors, MZF1 and E2F transcription factor 3 (E2F3), might regulate expression of these glycolytic genes ( Figure 1A). Among them, MZF1 was top transcription factor with six potential targets ( Figure S1) and chosen for further study. Notably, MZF1 was highly expressed in NB tissues with elder age (P=8.1×10 -5 ), death (P=1.3×10 -6 ), or advanced INSS stage (P=2.1×10 -4 ), and was associated with poor survival of patients (P=6.9×10 -4 and P=5.8×10 -3 , Figure 1B-C, and Figure S2A) as an independent prognostic factor (hazard ratio=2.330, 95% confidence interval=1.021 to 3.317, P=0.044). In addition, higher MZF1 expression was observed in NB cell lines, than that in normal dorsal root ganglia ( Figure S2B). To further elucidate the effects of MZF1 on glycolytic gene expression, we chose SH-SY5Y, SK-N-AS, BE(2)-C, and IMR-32 (with low and high MZF1 levels, respectively) cells as models. Sable over-expression or knockdown of MZF1 increased and decreased the levels of HK2 or PGK1, but not of fructose-bisphosphate C (ALDOC), ENO1, glucose-6-phosphate isomerase (GPI), or LDHA, in these NB cells, respectively ( Figure 1D-E and Figure S2C). The MZF1 enrichment on promoters of HK2 and PGK1 was increased and decreased by stable over-expression or knockdown of MZF1, respectively ( Figure 1F). Ectopic expression or knockdown of MZF1 facilitated and inhibited the promoter activity of HK2 and PGK1 in SH-SY5Y, SK-N-AS, BE(2)-C, and IMR-32 cells, respectively, while mutation of MZF1 binding site abolished these effects ( Figure 1G and Figure S2D). Consistently, mining of public datasets (GSE16476 and GSE62564) revealed that HK2 (P=3.0×10 -10 and P=2.4×10 -16 ) or PGK1 (P=5.3×10 -8 and P=3.0×10 -23 ) levels were associated with poor survival of NB patients ( Figure S2E), and were positively correlated with those of MZF1 (R=0.498, P=7.9×10 -7 ; R=0.408, P=8.0×10 -5 ; Figure S2F). High expression of MZF1, HK2, or PGK1 was also associated with poor survival of patients with breast cancer, endometrial carcinoma, glioma, head and neck carcinoma, lung cancer, lymphoma, pancreatic cancer, or renal clear cell carcinoma ( Figure S3). These data indicated that transcription factor MZF1 facilitated the expression of glycolytic genes in NB.

MZF1 promotes NB progression via facilitating aerobic glycolysis
To characterize the functional roles of MZF1 in NB cells, we applied dCas9-based clustered regularly interspaced short palindromic repeats (CRISPR) [25] to activate or repress expression of MZF1. As shown in Figure 2A, transfection of two independent dCas9a-MZF1 or dCas9i-MZF1 resulted in efficient over-expression or silencing of MZF1 in NB cells, respectively. Stable transfection-or dCas9a-induced up-regulation of MZF1 increased the ECAR, an indicator of glycolysis, in SH-SY5Y and SK-N-AS cells, while shRNA-or dCas9i-induced knockdown of MZF1 significantly attenuated glycolytic process in BE(2)-C and IMR-32 cells ( Figure 2B and Figure S4A). Meanwhile, OCR was reduced and enhanced in NB cells with over-expression or knockdown of MZF1, respectively ( Figure S4B). Accordingly, ectopic expression or knockdown of MZF1 increased and decreased the glucose uptake, lactate production, and ATP levels in NB cells, suggesting facilitated and reduced glycolysis, respectively (Figure 2C-D and Figure S4C-D).
To explore the roles of HK2 and PGK1 in MZF1-facilitated aerobic glycolysis, shRNAs or expression vectors of HK2 and PGK1 were transfected into SH-SY5Y and BE(2)-C cells to restore their expression, glucose uptake, lactate production, and ATP levels altered by stable over-expression or knockdown of MZF1 ( Figure S5A-B). In soft agar and matrigel invasion assays, the anchorage-independent growth and invasion of SH-SY5Y and BE(2)-C cells were enhanced and reduced by stable ectopic expression or knockdown of MZF1, which was partially rescued by silencing or over-expression of HK2 and PGK1, respectively ( Figure S6A-B). In addition, treatment with glycolysis inhibitor (2-DG) [26] abolished the increase in glucose uptake, lactate production, ATP levels, growth and invasion of NB cells induced by stable MZF1 over-expression ( Figure  S5B and Figure S6A-B). In vivo experiments using xenograft models revealed that stable over-expression of MZF1 promoted the tumorigenecity of SH-SY5Y cells, as displayed by increase in tumor growth, tumor weight, Ki-67 proliferative index, and elevated levels of HK2 and PGK1 ( Figure S7A). In contrast, stable silencing of MZF1 into BE(2)-C cells decreased the growth, weight, Ki-67 proliferation index, and expression levels of HK2 and PGK1 of subcutaneous xenograft tumors in nude mice ( Figure S7B).  In experimental metastasis assay, nude mice treated with tail vein injection of SH-SY5Y cells stably over-expressing MZF1 presented more lung metastatic counts and lower survival possibility ( Figure S7C), while stable knockdown of MZF1 into BE(2)-C cells resulted in less lung metastatic colonies and greater survival probability in nude mice ( Figure  S7D). Moreover, administration of 2-DG prevented the increased tumorigenesis and aggressiveness of NB cells in vivo induced by stable MZF1 over-expression ( Figure S7A and Figure S7C). These results suggested that MZF1 promoted tumor progression via facilitating aerobic glycolysis in NB.

MZF1-uORF-encoded peptide inhibits MZF1 expression
To explore self-regulatory mechanisms underlying MZF1 expression, we analyzed its 5'-UTR using ORF finder program (https://www.ncbi. nlm.nih.gov/orffinder), which revealed the existence of an uORF within this region ( Figure 3A). Insertion of 5'-UTR of MZF1 resulted in decrease of luciferase activity, while 5'-UTR along with MZF1 promoter fragment facilitated the activity of luciferase reporter ( Figure 3A). In addition, initiation codon mutation or frame-shift deletion of uORF within 5'-UTR led to significant increase in luciferase activity ( Figure 3A). Transfection of MZF1 CDS, but not MZF1 cDNA containing 5'-UTR, resulted in increased protein and transcript levels of MZF1 in SH-SY5Y cells ( Figure 3B and Figure S8A). Initiation codon mutation or frame-shift deletion of uORF resulted in increase of MZF1 levels in NB cells transfected by MZF1 cDNA containing 5'-UTR, while over-expression or knockdown of uORF led to decrease and increase in transcript and protein levels of MZF1, respectively ( Figure 3B and Figure S8A-B). Further mining of SmPort [27] and GWIPS-viz [28] databases implicated that this ribosome-binding uORF might encode a 21-amino acid peptide with high conservation in primates ( Figure 3C). Western blot assay using anti-GFP antibody indicated the fusion expression of uPEP with GFP in HEK293 cells, which was abolished by initiation codon mutation or frame-shift deletion of uORF ( Figure 3D). The translation of uPEP-GFP protein in BE(2)-C cells was also validated by Coomassie blue staining and western blot using a rabbit polyclonal antibody against uPEP ( Figure 3E and Figure S8C) and mutation of Kozak motif locating at upstream of GFP ( Figure 3F). Ectopic expression of GFP-tagged or Flag-tagged uPEP resulted in obvious decrease of MZF1 levels ( Figure 3F). Notably, in response to IGF1 stimulation, the distribution of uORF within heavy polysomes (within fractions 10-12) was decreased, while MZF1 antisense RNA 1 (MZF1-AS1), a noncoding transcript containing complementary sequence of uORF, was not enriched in heavy polysomes ( Figure S8D). Moreover, treatment with established glycolysis activator IGF1 [29] led to phosphorylation of AKT, down-regulation of MZF1-uPEP, and up-regulation of MZF1 in BE(2)-C cells, which was abolished by phosphatidylinositol 3 kinase inhibitor LY294002 ( Figure S8E). These data suggested that MZF1-uORF-encoded peptide inhibited MZF1 expression at transcriptional level in NB cells.

MZF1-uPEP interacts with YY1 to suppress its transactivation
To elucidate the mechanisms underlying MZF1-uPEP-inhibited MZF1 expression, we first observed its subcellular localization. As shown in Figure 4A-B, GST-tagged or Flag-tagged MZF1-uPEP was mainly expressed within the nuclei of HeLa and BE(2)-C cells. Immunofluorescence assay using MZF1-uPEP specific antibody also revealed the nuclear or cytoplasmic enrichment of MZF1-uPEP in BE(2)-C cells, which was enhanced by transfection of MZF1-uORF ( Figure 4C). Treatment with leptomycin B (LMB), an established nuclear export inhibitor [30], resulted in obvious aggregation of MZF1-uPEP within the nucleus of SH-SY5Y cells ( Figure 4C). Then, to identify the protein partner of MZF1-uPEP, we performed the co-IP followed by a proteomic analysis of pulled down proteins in BE(2)-C cells. Mass spectrometry revealed 1321 differential proteins between empty vector (mock) and MZF1-uORF transfection groups (Table S5), and two of them ( Figure 4D) were potential transcription factors regulating MZF1 expression revealed by UCSC Genome Browser (Table S6). Further validating co-IP and western blot assays indicated that YY1 protein, but not USF2, was able to interact with MZF1-uPEP in BE(2)-C cells, which was abolished by IGF1 treatment ( Figure 4E and Figure S8F). Co-localization of MZF1-uPEP and YY1 was observed in the nucleus of NB cells ( Figure 4F). Deletion-mapping experiments indicated that zinc finger (ZNF) domain of YY1 (amino acids 258-414) was required for its binding to MZF1-uPEP ( Figure 4G). Notably, the expression of YY1 and its interaction with MZF1-uPEP were higher in NB cell lines, than that in normal dorsal root ganglia ( Figure  S8G). Importantly, stable over-expression or knockdown of MZF1-uORF resulted in decreased and increased transactivation of YY1, which was prevented by ectopic expression or silencing of YY1 in BE(2)-C and SH-SY5Y cells, respectively ( Figure 4H).  In addition, transfection of YY1 into BE(2)-C cells led to increase in MZF1 promoter activity, while mutation of YY1 binding site or stable transfection of MZF1-uORF abolished these effects ( Figure S8H). Ectopic expression or knockdown of MZF1-uORF decreased and increased the binding of YY1 to MZF1 promoter in NB cells, which was abolished by stable over-expression or knockdown of YY1, respectively ( Figure 4I). Collectively, these results indicated that MZF1-uPEP interacted with YY1 to suppress its transactivation in NB cells.

MZF1-uPEP exerts tumor suppressive roles by repressing YY1
To further investigate the functional roles of MZF1-uPEP, we performed rescue studies in NB cells. Stable ectopic expression of MZF1-uORF prevented the increase in glucose uptake, lactate production, ATP levels, growth, invasion, and metastasis of NB cells in vitro and in vivo induced by IGF1 stimulation (Figure S9A-D). Stable transfection of MZF1-uORF or sh-uORF #1 led to significantly decreased and increased expression of MZF1 and downstream genes (HK2 and PGK1) in BE(2)-C and SH-SY5Y cells ( Figure  S10A-B), which was abolished by ectopic expression or knockdown of YY1, respectively ( Figure S10A-B). Meanwhile, ectopic expression or silencing of YY1 abolished the decrease or increase in glucose uptake, lactate production, and ATP levels of NB cells induced by stable over-expression or knockdown of MZF1-uORF ( Figure S10C). In MTT colorimetric, soft agar, and matrigel invasion assays, over-expression or silencing of YY1 reversed the decrease or increase of viability, growth and invasiveness of NB cells induced by stable ectopic expression or knockdown of MZF1-uORF, respectively (Figure S10D-F). These data indicated that MZF1-uPEP exerted tumor suppressive roles by repressing YY1.

Therapeutic efficiency of cell-penetrating MZF1-uPEP
Then, we further investigated the therapeutic efficiency of cell-penetrating MZF1-uPEP on biological behaviors of NB cells. Administration of a cell-penetrating FITC-labeled MZF1-uPEP with YY1 inhibiting properties, termed as YIP-21, resulted in its obvious nuclear enrichment in BE(2)-C cells ( Figure  5A). Biotin-labeled peptide pull-down assay revealed that YIP-21, but not control peptide (CTLP), was able to bind with YY1 ( Figure 5B). Administration of YIP-21 led to decrease in the viability of NB cells ( Figure 5C), but not of non-transformed MCF 10A or transformed HEK293 cells without endogenous interaction between MZF1-uPEP and YY1 ( Figure 5D and Figure S8G). In addition, treatment with YIP-21 decreased the anchorage-independent growth and invasion of BE(2)-C cells ( Figure 5E-F). To test in vivo therapeutic potency of YIP-21, tail vein administration of YIP-21 or CTLP was performed in nude mice bearing subcutaneous xenograft tumors or lung metastasis formed by BE(2)-C cells. Administration of YIP-21 resulted in decreased growth and weight of subcutaneous xenograft tumors in nude mice ( Figure  5G). The Ki-67-positive cells, expression of MZF1 and its downstream genes, glucose uptake, lactate production, and ATP levels within subcutaneous xenograft tumors were also significantly reduced by YIP-21 treatment (Figure 5G-H). In experimental metastasis assay, nude mice treated with YIP-21 presented with less lung metastatic counts and longer survival time ( Figure 5I). Consequently, these results demonstrated that cell-penetrating MZF1-uPEP suppressed tumorigenesis and aggressiveness of NB cells.

Therapeutic efficiency of MZF1-uORF over-expression
To further assess the therapeutic efficacy of lentivirus-mediated MZF1-uORF over-expression on tumor progression, nude mice were treated with subcutaneous or tail vein injection of IMR-32 cells stably expressing red fluorescent protein. One week later, mice were randomly divided into groups, and received intravenous administration of lentivirus carrying empty vector (mock) or MZF1-uORF. Administration of lentivirus-mediated MZF1-uORF dramatically reduced the growth and weight of xenograft tumors ( Figure S11A), decreased the Ki-67 proliferation index ( Figure S11A), increased the MZF1-uPEP levels ( Figure S11B), inhibited the expression of MZF1 and downstream glycolytic genes ( Figure S11C-D), and attenuated the glucose uptake, lactate production, and ATP levels within xenograft tumors ( Figure S11E). In experimental metastasis assay, nude mice treated with tail vein administration of lentivirus-mediated MZF1-uORF presented fewer lung metastatic counts and longer survival time ( Figure  S11F). These data indicated that lentivirus-mediated over-expression of MZF1-uORF suppressed NB progression.

Discussion
Aerobic glycolysis facilitates malignant cell transformation, tumor initiation and aggressive progression [5], while inhibition of glycolysis impairs growth and metastasis of many tumor cells [10,11], indicating an efficient therapeutic approach for tumors. Recent studies show that LDHA and LDHB are dispensable for aerobic glycolysis in NB [31], suggesting involvement of other glycolytic genes in this process. Among them, HK2 is mainly expressed in cancers, and phosphorylates glucose to produce glucose-6-phosphate, a rate-limiting and irreversible step of glycolysis [8]. In mouse models, HK2 plays a vital role in tumor initiation and maintenance [32]. Elevated HK2 is associated with poor survival of hepatocellular carcinoma, while inhibition of HK2 expression abrogates the tumorigenesis of tumor cells [33]. PGK1, a rate-limiting enzyme of glycolytic pathway, catalyzes the transfer of high-energy phosphate from 1-position of 1,3-diphosphoglycerate to ADP, and is essential for ATP generation [34]. PGK1 is up-regulated in breast cancer [35], pancreatic ductal adenocarcinoma [36], and hepatocellular carcinoma [37], while depletion of PGK1 dramatically reduces the proliferation and metastasis of cancer cells, indicating an oncogenic role of PGK1 in tumor progression [38]. In this study, we identify MZF1 as a transcription factor facilitating the expression of glycolytic genes HK2 and PGK1 in NB. In addition, we demonstrate that a peptide encoded by MZF1-uORF binds to YY1, resulting in decreased transactivation of YY1 and repressed expression of MZF1 and downstream glycolytic genes HK2 and PGK1 in NB cells (Figure 6G), implying a negative feedback loop of uORF-encoded peptide in MZF1 expression. Meanwhile, in response to stimulation of glycolysis activator, MZF1-uPEP is down-regulated to disrupt this negative feedback loop, resulting in enhanced MZF1 expression and aerobic glycolysis of NB cells.
MZF1, one member of Kruppel family proteins, is essential for the differentiation, proliferation, and migration of hematopoietic cells [39,40]. As a bi-functional transcription factor, MZF1 contains 13 zinc finger domains, and represses or activates gene transcription via binding to promoters [40]. Recent studies show that MZF1 plays an important role in tumorigenesis and aggressiveness. Forced expression of MZF1 induces malignant transformation of NIH3T3 cells, and initiates tumor formation in athymic mice [41]. MZF1 is involved in the etiology of many solid tumors, such as lung cancer [42], breast cancer [43], colorectal cancer [44], hepatocellular carcinoma [45], and cervical cancer [46]. MZF1 facilitates the transcription of c-MYC, and is responsible for growth, migration, and invasion of lung adenocarcinoma cells [42]. In breast cancer, MZF1 activates the expression of cathepsin B to increase the invasion of cancer cells [43]. Over-expression of MZF1 leads to transactivation of anexelekto (AXL) promoter and increase of migratory, invasive, and metastatic potential of colorectal cancer cells [44]. In hepatocellular carcinoma, MZF1 enhances the transcription of protein kinase C alpha (PKCα), thus facilitating the migration and invasion of cancer cells [45]. Meanwhile, MZF1 suppresses the migratory and invasive capability of cervical cancer cells by inhibiting transcription of matrix metalloproteinase-2 (MMP-2) [46]. These results indicate that MZF1 exerts oncogenic or tumor suppressive roles via transcriptional changes associated with malignant cell migration and invasiveness in a context-dependent manner. However, the roles of MZF1 in aerobic glycolysis during tumor progression still remain elusive. In this study, MZF1 was identified as an independent prognostic factor for poor outcome of NB patients, while HK2 and PGK1 were direct target genes of MZF1. Our gain-and loss-of-function studies indicated that MZF1 promoted aerobic glycolysis, growth, and invasiveness of NB cells, suggesting the oncogenic roles of MZF1 in NB progression.  The widespread presence of uORF within 5'-UTR is one of the mechanisms regulating gene expression [47,48]. Approximate 50% of human transcripts contain uORF [49], and uORF is able to repress translation of mRNAs through disturbing ribosomal scanning or altering mRNA stability [50]. For example, sex lethal protein binds to a cis-regulatory element within uORF, and imposes a negative effect on protein translation in Drosophila [51]. Recent ribosome profiling and validating studies indicate the generation of short peptides encoded by uORFs [47,50], and some uORF-encoded peptides are important for translational regulation [52,53]. The 5'-UTR of CCAAT/enhancer-binding protein homologous protein (CHOP) contains a conserved uORF which encodes a 31-amino acid peptide that inhibits the translation of CHOP [54]. In this study, we identified a conserved uORF within MZF1 5'-UTR, which encoded a small peptide that bound to YY1 protein.
Notably, MZF1-uPEP inhibited YY1-facilitated transcription of MZF1, indicting a novel action mode of uORF-encoded peptide in regulating gene transcription rather than protein translation. In addition, tumor suppressive functions of MZF1-uPEP were mediated, at least in part, through interacting with YY1 protein in NB cells.
YY1 is a transcription factor of GLI-Kruppel family, and plays a regulatory role in cellular growth, oncogenic transformation, epithelial-mesenchymal transition, and metastasis [55]. Human YY1 protein possesses a transactivation domain, a repression domain, and four C2H2-type zinc fingers [55], and activates or inactivates gene transcription depending on promoter contexts [56]. YY1 is highly expressed in many types of cancerous tissues, including prostate cancer, colon cancer, liver cancer, and lung cancer [57]. In colon cancer, YY1 promotes the growth and Wnt signaling pathway of cancer cells through inhibiting p53 [58]. In addition, YY1 facilitates the transcription of p-glycoprotein in acute lymphoblastic lekeumia, and is associated with poor survival of patients [59]. In this study, we found that YY1 promoted the expression of MZF1 in NB cells, resulting in facilitated glycolytic gene expression and tumor progression. In addition, MZF1-uPEP bound to zinc finger domain of YY1, resulting in repression of YY1 transactivation in NB cells. Importantly, administration of a cell-penetrating MZF1-uPEP or lentivirus over-expressing MZF1-uPEP was able to suppress aerobic glycolysis, tumorigenesis, and aggressiveness of NB cells, suggesting the crucial roles of MZF1-uPEP in repressing YY1/MZF1 axis in aerobic glycolysis and tumor progression.

Conclusions
In summary, we demonstrate that MZF1 is associated with poor outcome of NB, and exerts oncogenic roles in aerobic glycolysis and tumor progression.
Meanwhile, MZF1-uORF-encoded peptide suppresses the MZF1 expression, aerobic glycolysis, growth, and aggressiveness of NB cells. Mechanistically, MZF1 promotes the expression of glycolytic genes HK2 and PGK1, while MZF1-uPEP binds to YY1 to repress its transactivation, resulting in transcriptional suppression of MZF1 and downstream glycolytic genes. Administration of a cell-penetrating MZF1-uPEP or lentivirus over-expressing MZF1-uPEP suppresses the aerobic glycolysis, tumorigenesis, and aggressiveness of NB cells. Since MZF1 expression is negatively regulated by microRNAs let-7e and let-7d in breast cancer cells [60], the roles of let-7 family members in regulating MZF1-mediated aerobic glycolysis during NB progression warrant investigation. In addition, further studies are needed to explore the potential roles of noncoding RNA MZF1-AS1 in regulating MZF1-uPEP expression in NB. We believe that this study extends our knowledge about the regulation of aerobic glycolysis by transcription factor and its derived uPEP, and suggests that MZF1 and YY1 may be potential therapeutic targets for tumor progression.