Identification of Differential Genes Expression Profiles and Pathways of Bone Marrow Mesenchymal Stem Cells of Adolescent Idiopa

BASIC SCIENCE

Identification of Differential Genes Expression Profiles and Pathways of Bone Marrow Mesenchymal Stem Cells of Adolescent Idiopathic Scoliosis Patients by Microarray and Integrated Gene Network Analysis

Qianyu Zhuang, MD, Wenzhe Mao, PhD,y Pengchao Xu, PhD,y Hongling Li, PhD,y Zhao Sun, PhD,y Shugang Li, PhD, Guixing Qiu, MD, Jing Li, PhD,y and Jianguo Zhang, MD

transduction network analysis of DEGs contained in significant Study Design. Microarray approach and integrated gene net- pathways, 24 potential crucial genes were selected for validation work analysis. by reverse transcription polymerase chain reaction. Objective. To explore the differential genetic expression pro- Results. There are 1027 previously unrecognized DEGs in BM- file, gene ontology terms, and Kyoto Encyclopedia of Genes and MSCs from AIS patients. Pathway analysis revealed dysregulated Genomes pathways in bone marrow mesenchymal stem cells mitogen-activated protein kinase (MAPK) signaling pathway, PI3K- (BM-MSCs) of idiopathic scoliosis (AIS) and non-AIS controls. Akt signaling pathway, calcium signaling pathway, peroxisome Summary of Background Data. The pathogenesis of adoles- proliferator-activated receptor (PPAR) signaling pathway, ubiquitin- cent AIS and the accompanying generalized osteopenia remain mediated proteolysis, and Notch signaling pathway, all of which unclear. Our previous study suggested increased proliferation have been reported to play an important role in regulating the ability and decreased osteogenic differentiation ability of BM- osteogenic or adipogenic differentiation of MSCs. Furthermore, MSCs of AIS. Therefore, we hypothesized that MSCs may play a gene signal transduction networks analysis indicated that mitogen- significant role in the etiology and pathogenesis of AIS. Methods. In this study, microarray analysis was used to identify activated protein kinase kinase 1 (MAP2K1), SMAD family member differentially expressed genes (DEGs) of BM-MSCs from AIS 3 (SMAD3), homeobox C6 (HOXC6), heat shock 70kDa protein 6 patients compared with those from healthy individuals. Compre- (HSPA6), general transcription factor IIi (GTF2I), CREB binding hensive bioinformatics analyses were then used to enrich datasets protein (CREBBP), phosphoinositide-3-kinase, regulatory subunit 2 for gene ontology and pathway. Based on the gene signal (PIK3R2), and dual specificity phosphatase 2 (DUSP2) may play essential roles in AIS pathogenesis and accompanied osteopenia. Conclusion. This study reports the differential genes expres- From the Department of Orthopedics, Peking Union Medical College sion profiles of BM-MSCs from AIS patients and related potential y Hospital, Beijing, P.R. China; and Center of Excellence in Tissue Engin- pathways for the first time. These previously unrecognized genes eering, Institute of Basic Medical Sciences and School of Basic Medicine, Chinese Academy of Medical Sciences and Peking Union Medical College, and molecular pathways might play a significant role in not only Beijing, P.R. China. the causal mechanism of osteopenia in AIS, but also the AIS Acknowledgment date: June 30, 2015. First revision date: October 16, initiation and development. The identification of these candidate 2015. Acceptance date: November 2, 2015. genes provides novel insight into the underlying etiological Qianyu Zhuang and Wenzhe Mao contributed equally to this paper as the mechanisms of AIS. first authors. Key words: adolescent idiopathic scoliosis, bone marrow Jing Li and Jianguo Zhang contributed equally to this paper as the corre- mesenchymal stem cells, differentially expressed gene, gene sponding authors. expression profiles, microarray, osteopenia. The manuscript submitted does not contain information about medical Level of Evidence: N/A. device(s)/drug(s). Spine 2016;41:840–855 The National Natural Science Foundation of China (81272054, 81171673) funds were received to support this work. Relevant financial activities outside the submitted work: board membership, consultancy. Address correspondence and reprint requests to Jianguo Zhang, MD and tiology of adolescent idiopathic scoliosis (AIS) Professor, Peking Union Medical College Hospital, 1 Shuai Fu Yuan Beijing remains unclear.1–3 Low bone mineral density in 100730 P. R. China; E-mail: [email protected] E AIS reported by many authors suggests the possibility DOI: 10.1097/BRS.0000000000001394 of bone metabolism disturbance as an underlying 840 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

mechanism.4–8 In spite of the controversy engendered by for the following experiments. Primary antibodies against extensive hypotheses,9–12 there is a growing consensus that human CD29, CD31, CD34, CD44, CD45, CD73, and anomalies of bone growth and development are strongly CD105 (BD Biosciences) were used for immunophenotype related to the onset and progression of scoliosis.13-17 analysis of BM-MSCs. Mesenchymal stem cells (MSCs) act as progenitors of To identify the MSC capacity for multilineage differen- osteoblasts and regulate osteoclastogenesis via Receptor Acti- tiation, MSCs were cultured under differentiation con- vator for Nuclear Factor-k B Ligand and osteoprotegerin ditions. As previously described,21 cells were stained with expression.18–20 In addition, MSCs are indispensable in both the ALP staining kit (Beyotime, China) to reveal osteogenic intramembranous and endochondral bone formation.20 differentiation and stained with fresh Oil Red O solution Given the functional characteristics of MSCs in bone for- (Sigma, St.Louis, MO, USA) to show lipid droplets in mation and resorption, we hypothesized that MSCs play a induced cells. significant role in the etiology and pathogenesis of AIS. Our previous study21 identified 25 differentially Total RNA Extraction and Microarray Assay expressed proteins in MSCs from AIS patients using two- RNA was extracted from MSCs using TRIzol Reagent dimensional differential gel electrophoresis (2D-DIGE) and (Invitrogen, Carlsbad, CA, USA). Following purification MS-based proteomic approaches, and the function analysis with an RNeasy kit (Qiagen, Valencia, CA), cDNA was of these proteins suggested increased proliferation ability of generated using One-Cycle Target Labeling and Control MSCs and decreased osteogenic differentiation ability in Reagents (Affymetrix, Santa Clara, CA), and cRNA was AIS. These results were further supported by Park et al’s created with a GeneChip IVT Labeling Kit (Affymetrix, study,22 which revealed lower osteogenic differentiation Santa Clara, CA). Biotin-labeled, fragmented (200 nt) abilities and alkaline phosphatase (ALP) activities of MSCs cRNA was then hybridized for 16 hours at 458C to Affy- from AIS patients, indicating that the decreased osteogenic metrix GeneChip HumanTranscript 2.0 arrays (Affyme- differentiation ability of MSCs might be a possible mech- trix). GeneChips were washed and stained in the anism leading to low bone mass in AIS. In addition, a recent Affymetrix Fluidics Station 450. GeneChips were scanned study disclosed that the adipogenic ability of MSCs from by using Affymetrix GeneChip Command Console (AGCC) AIS girls was lower than controls.23 which installed in GeneChip Scanner 3000 7G. The data Based on these findings, we used a microarray approach were analyzed with robust multichip analysis algorithm in this study to further investigate specific alterations in the using Affymetrix default analysis settings and global scaling genetic expression profile of MSCs from AIS patients. as normalization method. Values presented are log2 robust Furthermore, the gene expression data were processed by multichip analysis signal intensity. gene ontology (GO), Kyoto Encyclopedia of Genes and Differentially expressed genes (DEGs) were identified Genomes (KEGG) orthology, and Signal network, which based on random-variance model t test and false discovery are effective bioinformatics analytical methods. rate (FDR) analysis. P < 0.05 and FDR < 0.05 was set as a threshold. Cluster 3.0 and TreeView analysis (Stanford MATERIALS AND METHODS University, California, USA) were performed to generate a dendrogram for each cluster of genes based on their Patients and Specimens expression profiling similarities. Bone marrow (BM) aspirates were obtained from 10 AIS patients (mean age 14.3 yr, range 12–17) and five non-AIS GO and Pathway Analysis patients with lower-leg fracture (mean age 14.6 yr, range Functional analysis of DEGs was carried out by the GO 12–17) (Table 1). In the AIS group, all of the patients under- project (http://www.geneontology.org) on the basis of bio- went full clinical and radiological examinations to rule out logical process [31], while pathway analysis was used to find other causes of scoliosis and to ascertain the diagnosis of out the significant pathway of the differential genes accord- AIS.24,25 In the control group, each of the five age- and sex- ing to KEGG (http://www.genome.jp/kegg/). The Fisher matched subjects had a straight spine and a normal forward exact test and x2 test were used to classify the GO category bending test on the physical examination. They were con- and pathway, and the FDR was calculated to correct the P firmed to be free of any associated medical diseases or spinal value. P < 0.05 and FDR < 0.1 were used as a threshold to deformities when entered to the study. The study was select significant GO categories and pathways. approved by the Ethics Committee of Peking Union Medical College Hospital. Written informed consents were obtained Signal Transduction Networks Analysis from all subjects and their parents before entering the study. Gene signal transduction network analysis (Signal-net), Isolation and Culture, Immunophenotype Analysis, based on KEGG database about the interactions between Osteogenic and Adipogenic Differentiation different gene products and the theory of network biology) of BM-MSCs was constructed based on the data of DEGs in significant As described in our previous study,21 BM-MSCs were iso- pathways. The ‘‘degree’’ is defined as the number of inter- lated using the same techniques and confluent cells actions of a gene with other genes in the gene network. (approximately 2 106) at the third passage were used Genes with higher degrees occupied more central positions Spine www.spinejournal.com 841 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

TABLE 1. Clinical Characteristics of AIS Patients Age at Lenke PUMC Location of Cobb Apical Spinal Patient Sex Diagnosis Classification24 Classification25 Major Curve Angle Vertebra Surgery Bracing 1 Female 13 5C- II D1 Left lumbar 418 L1 Yes No 2 Male 12 3BN II C3 Right thoracic/ 468/408 T8/L2 Yes Yes left lumbar 3 Male 16 1A- II B1 Right thoracic 508 T8 Yes No 4 Female 13 5CN I C Left lumbar 428 L1 Yes No 5 Female 17 1B- II B1 Right thoracic 578 T8 Yes Yes 6 Male 15 5CN I C Left lumbar 408 L2 Yes Yes 7 Female 14 6CN II C3 Right thoracic/ 408/448 T7/L1 Yes Yes left lumbar 8 Female 16 3B- II B1 Right thoracic/ 448/348 T8/L2 Yes No left lumbar 9 Male 14 3B- II B1 Right thoracic/ 448/348 T9/L2 Yes No left lumbar 10 Female 13 6CN II D2 Right thoracic/ 448/608 T6/L1 Yes Yes left lumbar The table presents characteristics of 10 patients with AIS who were selected based on described criteria. AIS indicates adolescent idiopathic scoliosis; PUMC, Peking Union Medical College.

in the network and had a stronger capacity of modulating negative for CD31, CD34, and CD45, but expressed high adjacent genes. levels of CD29, CD44, CD73, and CD105 (Supplementary Figure 1, http://links.lww.com/BRS/B80). Furthermore, they Validation of Microarray Data by Real-Time PCR can be differentiated into osteoblasts and adipocytes, which Twenty-four genes (MAP2K1, SMAD3, HOXC6, HSPA6, were verified by ALP and Oil Red O staining (Supplementary GTF2I, CREBBP, PIK3R2, DUSP2, etc) were selected for Figure 2, http://links.lww.com/BRS/B80). validation based both on their positions in signal network and their potential biological functions. Total RNA was isolated Microarray Gene Expression from the third passage of BM-MSCs from 10 AIS patients and An unsupervised hierarchical clustering of the microarray five controls. RNA was reverse transcribed using ReverTra data (Figure 1) revealed significant gene expression changes Ace qPCR RT Kit (TOYOBO, Osaka, Japan). Quantitative (random-variance model t test P < 0.05 and FDR P < 0.05) Real-time reverse transcription polymerase chain reaction in 1027 genes in AIS MSCs, as compared to controls (qRT-PCR) was conducted in 20-mL reactions consisting of (Supplementary Table 1, http://links.lww.com/BRS/B80). 10-mL SYBR Green Real-time PCR Master Mix, 1-mL10-mM Among these 1027 genes, 551 were upregulated (such as forward primers, 1-mL 10-mM reverse primers (listed in SMAD3, HOXC6, HOXC9, GTF2I, CREBBP, PIK3R2, Table 2), 1-mL template cDNA, and 7-mL double-distilled DUSP2, platelet-derived growth factor receptor, alpha poly- water. The PCR cycling programs began with an initial peptide PDGFRA) and 476 were downregulated (such as denaturation for 5 minutes at 948C, followed by 40 cycles MAP2K1, heat shock 70kDa protein 5, phosphoglycerate of 30 seconds at 948C, 30 seconds at 608C–658C, 30 seconds kinase 1, HSPA6, and ribophorin I). at 728C, and ended with the step of melting curve. The relative changes in gene expression were calculated by 244CT GO Analysis and Pathway Analysis Based on DEGs method; glyceraldehyde-3-phosphate dehydrogenase was In our study, the main GO categories for upregulated genes used as an internal control gene to normalize the amount were related to functions such as small GTPase-mediated of RNA added to the PCR reactions.50 signal transduction, DNA-dependent transcription, cytokinesis, completion of separation, etc (Figure 2A). The main RESULTS GO categories for downregulated genes were related to functions such as small molecule metabolic process, cell The Biological Characteristics of MSCs adhesion, transmembrane transport, immune response, All of the cultured cells from AIS patients and controls grew etc (Figure 2B). The differential genes within the significant well and displayed fibroblast-like morphology when GO categories are shown in Supplementary Table 2, http:// observed under a light microscope. To verify that these links.lww.com/BRS/B80. isolated cells were MSCs, we investigated their immunophe- Biological pathways analysis yielded significant upregu- notypes and multilineage differentiation capacities. These lated pathways and downregulated pathways. Among them, cells from both the AIS and control groups were persistently calcium signaling pathway, ubiquitin-mediated proteolysis,

842 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

TABLE 2. qRT-PCR Primers Sequences for Validated Targets Gene Name Ref_Seq mRNA Forward Primer (5’–3’) Reverse Primer (3’–5’) ENO2 NM_001975 CGTTATTGGCATGGATGTTG CCCAGTGATGTATCGGGAAG PGK1 NM_000291 ATGGATGAGGTGGTGAAAGC CAGTGCTCACATGGCTGACT MAP2K1 NM_002755 TGGATGGAGGTTCTCTGGAT TAGGATGTTGGAGGGCTTGA HSPA5 NM_005347 TAGCGTATGGTGCTGCTGTC TGACACCTCCCACAGTTTCA SERPINE1 NM_000602 GCTGTGTGTGAGCAGTGGAC CATCTTTGTGCCCTACCCTCT TPM3 NM_001043351 AGTCCCTCCAACCCTCAGAC GCTACCACCCACAAAGATGC RPN1 NM_002950 CTGGATTATGGGCCTTTCAG GTCATGCTGGTGATGGTCAG HSPA6 NM_002155 AGCAGACCCAGACTTTCACC AGCAGGTTGTTGTCCTTGG ERBB2 ENST00000540147 TGCTGTCCTGTTCACCACTC TGCTTTGCCACCATTCATTA SMARCC2 NM_001130420 GGCAAGGATGAGGATGAGAA TGATGATGTGGTGGGTCTGT PIK3R2 NM_005027 GATGGGCACTATGGCTTCTC TCCAGCTTGGCATTGTACTG SIK2 NM_015191 AGGTCAATGGCTGTCTGCTT GCCTCTCCTTCTGTCTCCAG HOXC9 NM_006897 GCTGGAACTGGAGAAGGAGTT GGCTGGGTAGGGTTTAGGAC CREBBP NM_001079846 CACAAACCCACTGATGAACG GTGCCAGCCTTTCCTTACAC CDK19 NM_015076 CGGCTTGTAGAGAGATTGCAC CAGCAGCCATACCTTCCTGT GTF2I NM_001163636 GGATGGTGGTGACATTCCTC TCAGTTCCGACGACAAACAC SMAD3 NM_001145102 CTGAAGCGCACTGACCATAA ATGCAGTTGTCCCATCCTGT HBP1 NM_012257 TTTGCCATCTTCACCTGGAT CATGCCAGATTGGGTAGGAT KAT2B NM_003884 GCCGTGTTATTGGTGGTATCT CCATAGCCCTTGACTTGCTC PDGFRA NM_006206 CGGTCTTGGAAGTGAGCAGT TGTAAATGTGCCTGCCTTCA DDB2 NM_000107 TTCTGGCATCAGTTCGCTTA ACTTCCGTGTCCTGGCTTC DUSP2 NM_004418 AACCAAGGGTGTGTCTGCTC CCAAGGGCTTCAACATGG DUSP6 NM_001946 ACAACAGGGTTCCAGCACA CAGACACATTCCAGCAAGGA HOXC6 NM_004503 CGCACAACTCTCTTTCACCA TCACTTGGAGGGCAATCTGT mRNA indicates messenger ribonucleic acid; qRT-PCR, quantitative reverse transcription polymerase chain reaction.

and Notch signaling pathway were upregulated (Figure 3A), 2D-DIGE and MS-based proteomic approaches.21 There- while MAPK signaling pathway, phosphatidylinositol-3 fore, the differential expressed genes in the present study kinase (PI3K)-Akt signaling pathway, PPAR signaling path- were matched with these previously recognized dysregu- way were downregulated (Figure 3B). The DEGs within lated proteins. Seventeen differential genes in this study significant pathways are shown in Supplementary Table 3, encode proteins that belong to the same family as the http://links.lww.com/BRS/B80. differential proteins in our previous research, which indicated that these genes required further concern and Signal Network Analysis of DEGs From Significant analysis (Table 3). Pathways Gene signal transduction network was constructed based on Profiled Genes Versus Previously Reported AIS the data of DEGs contained in significant pathways Candidate Regions (Figure 4) (Supplementary Table 4, http://links.lww.com/ Several chromosomal loci predisposing to AIS have been BRS/B80). From the most significant central genes, a total of identified by genetic linkage and genome-wide association 26–33 24 genes (MAP2K1, SMAD3, HOXC6, HSPA6, GTF2I, studies. But to date, no genes have been clearly ident- CREBBP, PIK3R2, DUSP2, etc) were selected for further ified as causative in AIS. We therefore attempted to identify real-time PCR validation considering both their positions in if there were any significantly differentially regulated genes the network and their potential biological functions. These in AIS MSCs within the reported loci. A total of 84 genes genes might be of great importance to construction of the were identified as corresponding with previously reported gene-gene interaction network in AIS MSCs, and therefore AIS candidate loci (Table 4). to AIS pathogenesis and accompanied osteopenia. The gene- gene interaction network of these 24 genes was also pre- Validation of Microarray Data by qRT-PCR Analysis sented (Figure 5) To independently confirm the microarray results, 24 genes were chosen for real-time quantitative PCR validation in 10 Differential Genes Matched With Previously Recognized AIS patients and five controls. The changes in the expression Dysregulated Proteins of AIS MSCs levels of the validated genes were paralleled in microarray In our previous study, 25 differentially expressed and qRT-PCR experiments, thereby confirming our results proteins of MSCs from AIS patients were identified using (Figure 6A,B). Spine www.spinejournal.com 843 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 1. Clustering analysis for the selected DEGs. The horizontal axis represents the samples, and the left vertical axis represents the genes. Red and green represent upregulation and downregulation separately. DEGs indicates differentially expressed genes.

844 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 2. The significant GO categories of the differential expressed genes. A and B show significant GO categories targeted by upregulated and downregulated categories, respectively. The vertical axis is the GO category, and the horizontal axis is the lg P of GO categories. GO indicates gene ontology.

Spine www.spinejournal.com 845 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 3. The significant pathways of the DEGs. A and B show significant pathways targeted by upregulated and downregulated pathways, respectively. The vertical axis is the pathway category, and the horizontal axis is the lg P of pathways. DEGs indicates differentially expressed genes.

Our results suggest that 1027 previously unrecognized DISCUSSION genes were differentially expressed in MSCs from AIS In the present study, to investigate the molecular mechanism patients. These genes are involved in multiple biological of decreased osteogenic differentiation ability of MSCs and processes, including small GTPase-mediated signal trans- gain an insight into the pathogenesis of AIS, we employed duction, DNA-dependent transcription, cytokinesis, cell microarray approach and integrated gene network analysis adhesion, transmembrane transport, immune response, to explore the differential genetic expression profile, GO etc. Results from pathway analysis revealed dysregulated terms, and KEGG pathways in MSCs of AIS and non-AIS MAPK signaling pathway, PI3K-Akt signaling pathway, controls. To the best of our knowledge, this study is the first calcium signaling pathway, PPAR signaling pathway, ubiq- microarray research on AIS in the field of MSCs. uitin-mediated proteolysis, and Notch signaling pathway,

846 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 4. The gene signal transduction networks of DEGs contained in significant pathways. Red cycle nodes represent upregulated genes, and blue cycle nodes represent downregulated genes. The nodes in the network were connected when their corresponding encoded gene products were connected directly or indirectly by a linker gene in the interaction network. Genes with higher degrees are more centralized in the network and have a stronger capacity of modulating adjacent genes. DEGs indicates differentially expressed genes.

all of which have been reported to play important role in the MAP2K1 gene was demonstrated by our study down- MSC osteogenic or adipogenic differentiation.34–36 regulated in AIS MSCs, which encode a mitogen-activated Together, the analysis of GO and pathway enrichment of protein kinase. Many previous studies have shown that these DEGs have provided novel insight into the molecular MAP2K1 lies upstream of extracellular signal-regulated pathogenesis of AIS initiation and development. kinases (ERKs) and plays an important role in osteogenic Our signal network analysis of DEGs involved in signifi- differentiation of MSCs and bone formation in vivo.37,38 A cant pathways could construct the gene-gene interaction constitutively active MAP2K1 can induce osteoblast gene network from differential genes, and could find the central expression, while inhibition of MAPK signaling intermedi- genes with the highest degree. In this study, 24 potential ates blocked differentiation.39 In addition, MAPK acti- crucial genes, including MAP2K1, SMAD3, HOXC6, vation via transfection of cells with constitutively active HSPA6, GTF2I, CREBBP, PIK3R2, DUSP2, were selected MAP2K1 (MAP2K1-SP) could induce the RuntRelated- for validation considering both their positions in the TranscriptionFactor2 (RUNX2)-responsive osteocalcin network and potential biological functions, which may (OCN) gene while dominant-negative MAP2K1 was inhibi- play essential roles in AIS pathogenesis and accompanied tory.40 Furthermore, transgenic study revealed that domi- osteopenia. nant-negative MAP2K1 mice showed decreased skeletal size Spine www.spinejournal.com 847 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 5. This sub-Signal-net shows related gene- gene interaction of the 24 selected significant genes. The red nodes represent the upregulated genes and the blue nodes downregulated genes. The area of the nodes represents the degree. Interaction between the genes is shown as a(b) activation (binding/association); ex, expression; a(ind), activation (indirect effect). and calvarial mineralization while these parameters were suggested that HOX gene family may be of significance in increased in MAP2K1-SP mice.38 Taken together these data the etiology of AIS and thus worth of in-depth exploration. suggest that downregulated expression of MAP2K1 might HSPA6, which was downregulated in MSCs from AIS play a significant role in the osteopenia in AIS and the group in this study, encodes heat shock protein 70 (HSP70). development of the disease. It should be noted that HSP70, being consistent with this Another important upregulated gene in AIS MSCs enc- study, has also been demonstrated to be downregulated odes SMAD3, a transcriptional modulator activated by expressed in AIS MSCs in our previous comparative pro- transforming growth factor-b (TGF-b). It has been showed teomics research.21 Many studies have reported that HSP70 that SMAD3 activated by TGF-b inhibits the expression of was involved in proliferation, differentiation, and apoptosis Runx2 and other Runx2 target genes, resulting in inhibition of BMSCs, by correcting misfolded proteins or helping fold of osteoblast differentiation of MSCs.41,42 In vivo, osteo- newly synthesized proteins during physiological proc- blasts and chondrocytes from SMAD3/ mice are unable esses.50–52 Furthermore, HSP70 has been reported to acti- to slow the progression of terminal differentiation, which vate ERK1/2 pathways for controlling proliferation and results in premature chondrocyte hypertrophy and growth osteogenic differentiation of MSCs.53,54 Since our serial plate mineralization and consequent chondrodysplasia.43 studies have shown the downregulation of HSP70 at both These previous findings and our observation suggest that genetic and protein levels, it is highly likely that its lower increased SMAD3 expression might be responsible, at least expression might subsequently compromise the capacity of in part, for the reduction of osteogenic differentiation MSCs to cope with aberrant polypeptides or damaged ability of AIS MSCs and persistent general osteopenia of proteins caused by environmental pathogenic factors of AIS patients. AIS. In addition, HSP70, as a mediator of osteogenic differ- HOXC6 and homeobox C9 (HOXC9), which are mem- entiation, might be related to decreased differential ability of bers of the Homeobox (HOX) gene family, were both upre- MSCs and clinical osteopenia in AIS patients. gulated in AIS MSCs according to our results. Many previous Other notable dysregulated genes include GTF2I, experiments have shown that HOX genes play a critical role in CREBBP, PIK3R2, and DUSP2. GTF2I encodes vertebrate- global patterning of the vertebrate axial skeleton44,45 direct specific transcription factors TFII-I, which was reported to functional evidence for this role has been provided by play an inhibitory role in regulating genes that are essential in both gain- and loss-of-function experiments.46–48 Interest- osteogenesis.55 CREBBP encodes a nuclear protein that binds ingly, a recent microarray analysis of AIS compared with to cAMP-response element binding protein (CREB), which is non-AIS osteoblasts identified differential expression of upstream of Runx2. The CREB-Smad6-Runx2 axis was HOXA group genes (10, 11, and 13), of HOXB group genes reported to play an essential role in the impaired osteogenesis (2–8) and of HOXD (1–4), as the most up- and downregu- of BMSCs with the fibrous dysplasia phenotype.56 PIK3R2 lated genes.49 These findings and our results strongly encoded protein is a regulatory component of PI3K, and

848 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Copyright © 2016WoltersKluwer Health,Inc.Unauthorized reproductionofthis articleisprohibited. Spine B ASIC TABLE 3. Profiled Genes Matched With Dysregulated Proteins of AIS MSCs S

Gene mRNA Geometric Mean Geometric Mean of CIENCE Symbol Accession Description Type Fold-change (A/C) P FDR of Intensities in A Intensities in C Style

HSPA13 NM_006948 Heat shock protein 70 kDa family, Coding 0.81 0.007 0.0067 109.82 135.86 Down member 13 HSPA5 NM_005347 Heat shock 70 kDa protein 5 Coding 0.85 0.039 0.035 1136.66 1330.4 Down (glucose-regulated protein, 78 kDa) HSPA6 NM_002155 heat shock 70 kDa protein 6 Coding 0.92 0.041 0.044 30.67 33.22 Down (HSP70B’) FKBP15 NM_015258 FK506-binding protein 15, 133 kDa Coding 0.93 0.048 0.0537 68.35 73.8 Down ATP6V0E1 NM_003945 ATPase, Hþ transporting, lysosomal Coding 0.88 0.026 0.0312 834.73 953.28 Down 9 kDa, V0 subunit e1 ATP6V1B2 NM_001693 ATPase, Hþ transporting, lysosomal Coding 0.92 0.044 0.049 103.63 112.17 Down 56/58 kDa, V1 subunit B2 ENO2 NM_001975 Enolase 2 (g, neuronal) Coding 0.72 0.042 0.0431 32.63 45.03 Down SERPINE1 NM_000602 Serpin peptidase inhibitor, clade E Coding 0.85 0.049 0.0521 1258.37 1475.26 Down (nexin, plasminogen activator inhibitor type 1), member 1 ACTG2 NM_001199893 Actin, g 2, smooth muscle, enteric Coding 0.96 0.020 0.0127 6.27 6.56 Down ACTC1 NM_005159 Actin, a, cardiac muscle 1 Coding 0.94 0.037 0.0386 18.05 19.29 Down

ACTN3 NM_001104 Actinin, a 3 Coding 0.95 0.045 0.0469 25.09 26.31 Down Patients AIS of BM-MSCs of DEGs of Identification ACTN2 NM_001103 Actinin, a 2 Coding 0.97 0.008 0.0079 5.01 5.17 Down TCP11L2 NM_152772 t-Complex 11 (mouse)-like 2 Coding 1.31 0.023 0.0206 24.69 18.8 Up ATP5I NM_007100 ATP synthase, Hþ transporting, Coding 1.13 0.025 0.025 91.53 81.32 Up mitochondrial Fo complex, subunit E CAPG NM_001256139 Capping protein (actin filament), Coding 1.08 0.037 0.0342 44.22 40.88 Up gelsolin-like ANXA3 NM_005139 Annexin A3 Coding 1.02 0.043 0.0389 3 2.96 Up ANXA11 NM_001157 Annexin A11 Coding 1.07 0.046 0.0479 70.5 65.85 Up Seventeen differential genes in this study encode proteins that belong to the same family as the differential proteins in our previous research, which indicated that these genes required further concern and analysis. AIS-MSCs indicates adolescent idiopathic scoliosis-mesenchymal stem cells; FDR, false discovery rate; mRNA, messenger ribonucleic acid. www.spinejournal.com hage al et Zhuang 849 Copyright © 2016WoltersKluwer Health,Inc.Unauthorized reproductionofthis articleisprohibited. 850 TABLE 4. Transcriptome Profile of Genes Within Reported AIS Candidate Loci

www.spinejournal.com Candidate Region References Candidate Gene Ref_Seq Chromosomal Localization P FDR Fold-change (A/C)

19p13.3 Chan et al BTBD2 NM_017797 19p13.3 0.012 0.01 1.15 B 31 Alden et al ANGPTL4 NM_139314 19p13.3 0.019 0.02 0.6 ASIC PLEKHJ1 NM_018049 19p13.3 0.018 0.02 0.92 REXO1 NM_020695 19p13.3 0.049 0.05 0.96 S 33 12p13.31 Raggio et al A2M NM_000014 12p13.31 0.005 0.01 0.59 CIENCE SLC2A14 NM_153449 12p13.31 0.033 0.03 0.82 NECAP1 NM_015509 12p13.31 0.030 0.03 0.85 ING4 NM_001127582 12p13.31 0.010 0.01 1.12 17q25.3-qtel Ocaka et al30 WDR45L NM_019613 17q25.3 0.007 0.01 0.82 SLC38A10 NM_001037984 17q25.3 0.019 0.02 0.91 USP36 NM_025090 17q25.3 0.039 0.04 0.93 9q31.2-q34.2 Ocaka et al30 TMEM38B NM_018112 9q31.2 0.026 0.02 0.72 Xq23-26.1 Justice et al29 KLHL13 NM_001168299 Xq23-q24 0.043 0.04 1.04 NDUFA1 NM_004541 Xq23 0.009 0.01 1.15 LRCH2 NM_001243963 Xq23 0.010 0.01 1.19 CUL4B NM_001079872 Xq24 0.046 0.05 1.27 DOCK11 NM_144658 Xq26.1 0.010 0.01 1.32 ZDHHC9 NM_001008222 Xq26.1 0.015 0.01 0.85 17p11.2 Salehi et al28 LLGL1 NM_004140 17p11.2 0.015 0.01 1.09 5q13 Edery et al27 FAM169A NM_015566 5q13.3 0.024 0.02 1.07 CCDC125 NM_176816 5q13.2 0.023 0.02 1.11 SV2C BC100826 5q13.3 0.007 0.01 0.96 26 18q Wise et al NETO1 NM_001201465 18q22.2 0.015 0.02 0.95 Patients AIS of BM-MSCs of DEGs of Identification PSMA8 NM_144662 18q11.2 0.019 0.02 0.97 TMEM241 NM_032933 18q11.2 0.020 0.02 0.87 FECH NM_000140 18q21.3 0.005 0.00 1.15 ZSCAN30 NM_001112734 18q12.2 0.018 0.02 1.17 TSHZ1 NM_005786 18q22.3 0.004 0.00 1.21 DTNA NM_032975 18q12 0.008 0.01 1.23 10q Wise et al26 KCNMA1 NM_001014797 10q22.3 0.050 0.05 1.22 PLAU NM_001145031 10q24 0.025 0.03 1.37 BICC1 NM_001080512 10q21.1 0.004 0.00 1.45 ANKRD1 NM_014391 10q23.31 0.035 0.03 0.5 ACSL5 NM_016234 10q25.1-q25.2 0.023 0.02 0.7 FUT11 NM_173540 10q22.2 0.003 0.00 0.83 PPRC1 NM_015062 10q24.32 0.029 0.03 0.86 PLAC9 BC090922 10q22.3 0.013 0.02 0.87 EIF5AL1 NM_001099692 10q22.3 0.015 0.02 0.88 VDAC2 NM_001184783 10q22 0.025 0.02 0.89

LOC100132116 ENST00000437930 10q23.31 0.036 0.04 0.9 hage al et Zhuang FAM25C NM_001137548 10q11.22 0.041 0.04 0.91

a 2016 May GPRIN2 NM_014696 10q11.22 0.026 0.02 0.91 LRIT1 NM_015613 10q23 0.012 0.01 0.92 MBL2 NM_000242 10q11.2 0.003 0.00 0.93 ZNF239 NM_001099282 10q11.22-q11.23 0.043 0.02 0.93 Copyright © 2016WoltersKluwer Health,Inc.Unauthorized reproductionofthis articleisprohibited. Spine TABLE 4 (Continued)

Candidate Region References Candidate Gene Ref_Seq Chromosomal Localization P FDR Fold-change (A/C) PCDH15 NM_001142767 10q21.1 0.038 0.02 0.94 B

SLC18A2 NM_003054 10q25 0.005 0.01 0.96 ASIC LRRC27 NM_001143757 10q26.3 0.005 0.01 1.04 ARL3 NM_004311 10q23.3 0.047 0.05 1.05 S

TBC1D12 NM_015188 10q23.3 0.022 0.02 1.06 CIENCE MARCH8 NM_145021 10q11.21 0.007 0.01 1.07 ANXA11 NM_001157 10q23 0.047 0.05 1.07 MARVELD1 NM_031484 10q24.2 0.018 0.01 1.09 RASSF4 NM_032023 10q11.21 0.006 0.01 1.09 EIF4EBP2 NM_004096 10q21-q22 0.037 0.04 1.1 KIAA1598 NM_001127211 10q25.3 0.046 0.04 1.12 PRKG1 NM_001098512 10q11.2 0.007 0.01 1.19 6q Wise et al26 ARID1B NM_017519 6q25.1 0.015 0.01 1.08 PLN NM_002667 6q22.1 0.027 0.03 1.08 CCDC28A NM_015439 6q23.1-q24.1 0.033 0.04 1.12 GJB7 NM_198568 6q15 0.022 0.02 1.13 CTAGE9 NM_001145659 6q23.2 0.045 0.05 1.14 CDK19 NM_015076 6q21 0.050 0.05 1.15 TIAM2 NM_012454 6q25.2 0.023 0.03 1.15 SESN1 NM_001199933 6q21 0.021 0.02 1.17 PHACTR2 NM_014721 6q24.2 0.028 0.03 1.21 ELOVL4 NM_022726 6q14 0.006 0.00 1.33 SLC2A12 NM_145176 6q23.2 0.033 0.03 1.39 dniiaino Eso MMC fASPatients AIS of BM-MSCs of DEGs of Identification HECA NM_016217 6q23-q24 0.003 0.00 1.41 CD109 NM_001159587 6q13 0.038 0.04 1.42 COL12A1 NM_004370 6q12-q13 0.004 0.00 1.48 RPF2 NM_032194 6q21 0.013 0.01 0.79 SGK1 NM_001143676 6q23 0.040 0.03 0.79 TAAR6 NM_175067 6q23.2 0.006 0.01 0.89 SFT2D1 NM_145169 6q27 0.044 0.04 0.91 AIG1 NM_016108 6q24.2 0.007 0.01 0.92 MB21D1 NM_138441 6q13 0.043 0.03 0.92 SYTL3 NM_001009991 6q25.3 0.025 0.02 0.93 KLHL32 NM_052904 6q16.1 0.027 0.03 0.96 EPM2A NM_001018041 6q24 0.048 0.05 0.97 www.spinejournal.com TTLL2 NM_031949 6q27 0.023 0.02 0.97 PRDM1 NM_001198 6q21 0.040 0.04 0.98 BACH2 NM_001170794 6q15 0.005 0.00 1.03 FABP7 NM_001446 6q22-q23 0.013 0.01 1.04

The table presents a total of 84 genes within previously reported AIS candidate loci identified by genetic linkage and genome-wide association studies. AIS indicates adolescent idiopathic scoliosis; FDR, false discovery rate. al et Zhuang 851 BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

Figure 6. Validation of selected 24 downregulated A and upregulated B genes by qRT-PCR. The relative level of mRNA of regulated genes was analyzed by qRT-PCR. Gene expression results are depicted as 2DDCt values, normalized to GAPDH. P < 0.05, Student t test, AIS (black) versus control (white) expression levels. GAPDH indicates glyceraldehyde-3-phosphate dehydrogenase; mRNA, messenger ribonucleic acid; qRT-PCR, quantitative reverse transcription PCR.

852 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

PI3K/Akt signaling pathway and downstream targets have signaling pathway, ubiquitin-mediated proteolysis, been shown to be critical regulators of bone formation and and Notch signaling pathway were dysregulated in remodeling.35,57 DUSP2 gene product is reported to negatively AIS MSCs. regulate members of the mitogen-activated protein kinase MAP2K, SMAD3, HOXC6, HSPA6, GTF2I, CREBBP, superfamily, which play an important role in osteogenic differ- PIK3R2, DUSP2 may play essential roles in AIS entiation of MSCs and bone formation in vivo.58 pathogenesis and accompanied osteopenia. In addition, other significant dysregulated genes presented in Figure 6, such as K(lysine) acetyltransferase 2B (KAT2B), erythroblastic leukemia viral oncogene homolog 2 (ERBB2), DUSP6, tropomyosin 3 (TPM3), enolase 2 Acknowledgments (ENO2), and damage-specific DNA binding protein 2 Microarray data were analyzed by Genminix Informatics (DDB2), are also worth our concern. KAT2B encodes a (Shanghai, China). catalytic subunit for histones acetylation, which is an Supplemental digital content is available for this article. important regulator for promoting osteoblast differen- Direct URL citations appearing in the printed text are tiation via acetylation modification of Runx2.59,60 ERBB2 provided in the HTML and PDF version of this article on encodes a member of the epidermal growth factor receptor the journal’s Web site (www.spinejournal.com). family, which can enhance kinase-mediated activation of MAPK and PI3K signaling pathways.61,62 DUSP6 encodes a dual specificity phosphatase exclusively specific to MAPK1/ References 1. Altaf F, Gibson A, Dannawi Z, et al. Adolescent idiopathic ERK2, which is demonstrated to play a negative regulatory scoliosis. BMJ 2013;346:f2508. 63,64 role in MAPK1 in a feedback loop manner. TPM3 gene 2. Dayer R, Haumont T, Belaieff W, et al. Idiopathic scoliosis: product provides stability to actin filaments and regulates etiological concepts and hypotheses. J Child OrthopV 7 2013; access of actin-binding proteins. Interestingly, it was 11–6. 3. Cheung KM, Wang T, Qiu GX, et al. Recent advances in the revealed that disrupting actin in MSCs increased adipo- aetiology of adolescent idiopathic scoliosis. Int Orthop 2008;32: genesis and decreased osteogenesis when compared to 729–34. untreated controls, suggesting that the actin cytoskeleton 4. Yu WS, Chan KY, Yu FW, Yeung HY, et al. Abnormal bone 65 quality versus low bone mineral density in adolescent idiopathic might be important in the commitment process. ENO2 scoliosis: a case-control study with in vivo high-resolution periph- encodes one of the three enolase isoenzymes (a, b, g), while eral quantitative computed tomography. Spine J 2013;13:1493–9. a-enolase was reported to be downregulated expressed in 5. Pourabbas Tahvildari B, Erfani MA, Nouraei H, et al. Evaluation AIS MSCs in our previous proteomics research.21 DDB2- of bone mineral status in adolescent idiopathic scoliosis. Clin Orthop Surg 2014;6:180–4. encoded protein participates in nucleotide excision repair, 66,67 6. Li XF, Li H, Liu ZD, et al. Low bone mineral status in adolescent and facilitates the cellular response to DNA damage. idiopathic scoliosis. Eur Spine JV 17 2008;1431–40. The possible mechanism of the observed dysregulation of 7. Sadat-Ali M, Al-Othman A, Bubshait D, et al. Does scoliosis causes these genes expression in the etiology of AIS required further low bone mass? A comparative study between siblings. Eur Spine J 2008;17:944–7. research and elucidation. 8. Cheng JC, Hung VW, Lee WT, et al. Persistent osteopenia in In summary, we described the differential gene expres- adolescent idiopathic scoliosis—longitudinal monitoring of bone sion profiles of BM-MSCs from AIS patients and related mineral density until skeletal maturity. Stud Health Technol potential pathways for the first time. These previously Inform 2006;123:47–51. 9. Goultidis TT, Papavasiliou KA, Petropoulos AS, et al. Higher unrecognized genes and molecular pathways might play a levels of melatonin in early stages of adolescent idiopathic scolio- significant role, in not only the causal mechanism of osteo- sis: toward a new scenario. J Pediatr Orthop 2014;34:768–73. penia in AIS, but also the initiation and development of AIS. 10. Zhang Y, Gu Z, Qiu G. The association study of calmodulin 1 gene The identification of these candidate genes provides novel polymorphisms with susceptibility to adolescent idiopathic scoliosis. Biomed Res Int 2014;2014:168106. insight into the underlying etiological mechanisms of AIS. 11. Hawasli AH, Hullar TE, Dorward IG. Idiopathic scoliosis and the Further studies are required to clarify the underlying mech- vestibular system. Eur Spine J 2015;24:227–33. anism of MSCs in AIS pathogenesis. 12. Xu L, Huang S, Qin X, et al. Investigation of the 53 markers in a DNA-based prognostic test revealing new predisposition genes for adolescent idiopathic scoliosis. Spine 2015;40:1086–91. 13. Ishida K, Aota Y, Mitsugi N, et al. Relationship between bone Key Points density and bone metabolism in adolescent idiopathic scoliosis. Scoliosis 2015;10:9. Differential genes expression profiles of BM-MSCs 14. Kono H, Machida M, Saito M, et al. Mechanism of osteoporosis in from AIS patients and related potential pathways adolescent idiopathic scoliosis: experimental scoliosis in pinealec- were reported for the first time. tomized chickens. J Pineal Res 2011;51:387–93. 15. Cheung CS, Lee WT, Tse YK, et al. Generalized osteopenia in A total of 1027 previously unrecognized genes adolescent idiopathic scoliosis—association with abnormal puber- were differential expressed in MSCs from AIS tal growth, bone turnover, and calcium intake? Spine 2006;31: patients. 330–8. MAPK signaling pathway, PI3K-Akt signaling 16. Cheng JC, Tang SP, Guo X, et al. Osteopenia in adolescent idiopathic scoliosis: a histomorphometric study. Spine 2001;26: pathway, calcium signaling pathway, PPAR E19–23.

Spine www.spinejournal.com 853 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

17. Dede O, Akel I, Demirkiran G, et al. Is decreased bone mineral 40. Xiao G, Jiang D, Thomas P, et al. MAPK pathways activate and density associated with development of scoliosis? A bipedal osteo- phosphorylate the osteoblast-specific transcription factor, cbfa1. penic rat model. Scoliosis 2011;6:24. J Biol Chem 2000;275:4453–9. 18. Ekizer A, Yalvac ME, Uysal T, et al. Bone marrow mesenchymal 41. Kang JS, Alliston T, Delston R, et al. Repression of Runx2 function stem cells enhance bone formation in orthodontically expanded by TGF-beta through recruitment of class II histone deacetylases maxillae in rats. Angle Orthod 2015;85:394–9. by Smad3. EMBO J 2005;24:2543–55. 19. de Santana Santos T, Flores Abuna RP, Bacha Lopes H, et al. 42. Alliston T, Choy L, Ducy P, et al. TGF-beta-induced repression of Association of mesenchymal stem cells and osteoblasts for bone CBFA1 by Smad3 decreases CBFA1 and osteocalcin expression repair. Regen Med 2015;10:127–33. and inhibits osteoblast differentiation. EMBO J 2001;20: 20. Bielby R, Jones E, McGonagle D. The role of mesenchymal stem 2254–72. cells in maintenance and repair of bone. Injury 2007;38 (suppl 1): 43. Yang X, Chen L, Xu X, et al. TGF-beta/Smad3 signals repress S26–32. chondrocyte hypertrophic differentiation and are required for 21. Zhuang Q, Li J, Wu Z, et al. Differential proteome analysis of bone maintaining articular cartilage. J Cell Biol 2001;153:35–46. marrow mesenchymal stem cells from adolescent idiopathic sco- 44. Gaunt SJ. Evolutionary shifts of vertebrate structures and Hox liosis patients. PLoS One 2011;6:e18834. expression up and down the axial series of segments: a consider- 22. Park WW, Suh KT, Kim JI, et al. Decreased osteogenic differen- ation of possible mechanisms. Int J Dev Biol 2000;44: tiation of mesenchymal stem cells and reduced bone mineral 109–17. density in patients with adolescent idiopathic scoliosis. Eur Spine 45. Wellik DM. Hox genes and vertebrate axial pattern. Curr Top Dev J 2009;18:1920–6. Biol 2009;88:257–78. 23. Liang G, Gao W, Liang A, et al. Normal leptin expression, lower 46. Carapuco M, Novoa A, Bobola N, et al. Hox genes specify adipogenic ability, decreased leptin receptor and hyposensitivity to vertebral types in the presomitic mesoderm. Genes Dev leptin in adolescent idiopathic scoliosis. PLoS One 2012;7: 2005;19:2116–21. e36648. 47. Jegalian BG, De Robertis EM. Homeotic transformations in the 24. Lenke LG. The Lenke classification system of operative adolescent mouse induced by overexpression of a human Hox3.3 transgene. idiopathic scoliosis. Neurosurg Clin N Am 2007;18:199–206. Cell 1992;71:901–10. 25. Qiu G, Zhang J, Wang Y, et al. A new operative classification of 48. Condie BG, Capecchi MR. Mice with targeted disruptions in the idiopathic scoliosis: a peking union medical college method. Spine paralogous genes hoxa-3 and hoxd-3 reveal synergistic inter- 2005;30:1419–26. actions. Nature 1994;370:304–7. 26. Wise CA, Barnes R, Gillum J, et al. Localization of susceptibility to 49. Fendri K, Patten SA, Kaufman GN, et al. Microarray expression familial idiopathic scoliosis. Spine 2000;25:2372–80. profiling identifies genes with altered expression in adolescent 27. Edery P, Margaritte-Jeannin P, Biot B, et al. New disease gene idiopathic scoliosis. Eur Spine J 2013;22:1300–11. location and high genetic heterogeneity in idiopathic scoliosis. Eur 50. Chen J, Shi ZD, Ji X, et al. Enhanced osteogenesis of human J Hum Genet 2011;19:865–9. mesenchymal stem cells by periodic heat shock in self-assembling 28. Salehi LB, Mangino M, De Serio S, et al. Assignment of a locus for peptide hydrogel. Tissue Eng Part A 2013;19:716–28. autosomal dominant idiopathic scoliosis (is) to human chromo- 51. Mosser DD, Caron AW, Bourget L, et al. Role of the human heat some 17p11. Hum Genet 2002;111:401–4. shock protein hsp70 in protection against stress-induced apoptosis. 29. Justice CM, Miller NH, Marosy B, et al. Familial idiopathic Mol Cell Biol 1997;17:5317–27. scoliosis: evidence of an x-linked susceptibility locus. Spine 52. Frydman J, Hartl FU. Principles of chaperone-assisted protein 2003;28:589–94. folding: differences between in vitro and in vivo mechanisms. 30. Ocaka L, Zhao C, Reed JA, et al. Assignment of two loci for Science 1996;272:1497–502. autosomal dominant adolescent idiopathic scoliosis to chromo- 53. Hronik-Tupaj M, Rice WL, Cronin-Golomb M, et al. Osteoblastic somes 9q31.2-q34.2 and 17q25.3-qtel. J Med Genet 2008;45: differentiation and stress response of human mesenchymal stem 87–92. cells exposed to alternating current electric fields. Biomed Eng 31. Alden KJ, Marosy B, Nzegwu N, et al. Idiopathic scoliosis: Online 2011;10:9. identification of candidate regions on chromosome 19p13. Spine 54. Raucci A, Bellosta P, Grassi R, et al. Osteoblast proliferation or 2006;31:1815–9. differentiation is regulated by relative strengths of opposing signal- 32. Chan V, Fong GC, Luk KD, et al. A genetic locus for adolescent ing pathways. J Cell Physiol 2008;215:442–51. idiopathic scoliosis linked to chromosome 19p13.3. Am J Hum 55. Lazebnik MB, Tussie-Luna MI, Hinds PW, et al. Williams- Genet 2002;71:401–6. Beuren syndrome-associated transcription factor TFII-I 33. Raggio CL, Giampietro PF, Dobrin S, et al. A novel locus for regulates osteogenic marker genes. J Biol Chem 2009;284: adolescent idiopathic scoliosis on chromosome 12p. J Orthop Res 36234–9. 2009;27:1366–72. 56. Fan QM, Yue B, Bian ZY, et al. The CREB-Smad6-Runx2 axis 34. James AW. Review of signaling pathways governing MSC osteo- contributes to the impaired osteogenesis potential of bone mar- genic and adipogenic differentiation. Scientifica 2013;2013: row stromal cells in fibrous dysplasia of bone. JPathol2012;228: 684736. 45–55. 35. Chen J, Crawford R, Chen C, et al. The key regulatory roles of the 57. Tong Y, Feng W, Wu Y, et al. Mechano-growth factor accelerates PI3K/Akt signaling pathway in the functionalities of mesenchymal the proliferation and osteogenic differentiation of rabbit mesen- stem cells and applications in tissue regeneration. Tissue Eng Part chymal stem cells through the PI3K/AKT pathway. BMC Biochem B Rev 2013;19:516–28. 2015;16:1. 36. Yuan Z, Li Q, Luo S, et al. Ppargamma and wnt signaling in 58. Chappell J, Sun Y, Singh A, et al. MYC/MAX control ERK adipogenic and osteogenic differentiation of mesenchymal stem signaling and pluripotency by regulation of dual-specificity phos- cells. Curr Stem Cell Res Ther 2015; (Epub ahead of print). phatases 2 and 7. Genes Dev 2013;27:725–33. 37. Kanno T, Takahashi T, Tsujisawa T, et al. Mechanical stress- 59. Shi S, Lin J, Cai Y, et al. Dimeric structure of p300/CBP associated mediated Runx2 activation is dependent on Ras/ERK1/2 MAPK factor. BMC Struct Biol 2014;14:2. signaling in osteoblasts. J Cell Biochem 2007;101:1266–77. 60. Wang CY, Yang SF, Wang Z, et al. PCAF acetylates Runx2 and 38. Ge C, Xiao G, Jiang D, et al. Critical role of the extracellular promotes osteoblast differentiation. J Bone Miner Metab signal-regulated kinase-MAPK pathway in osteoblast differen- 2013;31:381–9. tiation and skeletal development. JCellBiol2007;176: 61. Sun KK, Zhong N, Yang Y, et al. Enhanced radiosensitivity of 709–18. NSCLC cells by transducer of erbB2.1 (TOB1) through modu- 39. Franceschi RT, Ge C, Xiao G, et al. Transcriptional regulation of lation of the MAPK/ERK pathway. Oncol Rep 2013;29: osteoblasts. Ann N Y Acad Sci 2007;1116:196–207. 2385–91.

854 www.spinejournal.com May 2016 Copyright © 2016 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. BASIC SCIENCE Identification of DEGs of BM-MSCs of AIS Patients Zhuang et al

62. Akhtar S, Yousif MH, Chandrasekhar B, et al. Activation of 65. Lawlor MW, Dechene ET, Roumm E, et al. Mutations of tropo- EGFR/ERBB2 via pathways involving ERK1/2, P38 MAPK, myosin 3 (TPM3) are common and associated with type 1 myo- AKT and FOXO enhances recovery of diabetic hearts from ische- fiber hypotrophy in congenital fiber type disproportion. Hum Mut mia-reperfusion injury. PLoS One 2012;7:e39066. 2010;31:176–83. 63. Zhang H, Guo Q, Wang C, et al. Dual-specificity phosphatase 6 66. Stoyanova T, Roy N, Kopanja D, et al. DDB2 (damaged-DNA (Dusp6), a negative regulator of FGF2/ERK1/2 signaling, enhances binding protein 2) in nucleotide excision repair and DNA damage 17beta-estradiol-induced cell growth in endometrial adenocarci- response. Cell Cycle 2009;8:4067–71. noma cell. Mol Cell Endocrinol 2013;376:60–9. 67. Fei J, Kaczmarek N, Luch A, et al. Regulation of nucleotide 64. Domercq M, Alberdi E, Sanchez-Gomez MV, et al. Dual-specific excision repair by UV-DDB: prioritization of damage phosphatase-6 (Dusp6) and ERK mediate AMPA receptor-induced recognition to internucleosomal DNA. PLoS Biol 2011;9: oligodendrocyte death. J Biol Chem 2011;286:11825–36. e1001183.