1 Genome and Epigenome

23 SATB2-AS1 SuppressesQ1 Colorectal Carcinoma4 Aggressiveness by Inhibiting SATB2-Dependent5 Snail Transcription and Epithelial–Mesenchymal6 TransitionQ2

7 Yi-Qing Wang1,2, Dong-Mei Jiang1,2, Sha-Sha Hu1,2, Li Zhao1,2, Lan Wang2, Min-hui Yang1,2,8 Mei-Ling Ai2, Hui-Juan Jiang2, Yue Han2, Yan-Qing Ding1,2, and Shuang Wang1,2

9 Abstract

11 Accumulating evidence suggests that long noncoding RNA12 (lncRNA) plays important regulatory roles in cancer biology.13 However, the involvement of lncRNA in colorectal carcinoma14 progression remains largely unknown, especially in colorectal15 carcinomaQ5 metastasis. In this study, we investigated the16 changes in lncRNA expression in colorectal carcinoma and17 identified a new lncRNA, the antisense transcript of SATB218 (SATB2-AS1), as a key regulator of colorectal carcinoma pro-19 gression. SATB2-AS1 was frequently downregulated in colo-20 rectal carcinoma cells and tissues, and patients whose tumors21 expressed SATB2-AS1 at low levels had a shorter overall22 survival and poorer prognosis. Downregulation of SATB2-AS123 significantly promoted cell proliferation, migration, and inva-24 sion in vitro and in vivo, demonstrating that it acts as a tumor

25 suppressor in colorectal carcinoma. SATB2-AS1 suppressed26 colorectal carcinoma progression by serving as a scaffold27 to recruit p300, whose acetylation of H3K27 and H3K9 at28 the SATB2 promoter upregulated expression of SATB2, a29 suppressor of colorectal carcinoma growth and metastasis.30 SATB2 subsequently recruited HDAC1 to the Snail promoter,31 repressing Snail transcription and inhibiting epithelial-to-32 mesenchymal transition. Taken together, these data reveal33 SATB2-AS1 as a novel regulator of the SATB2-Snail axis whose34 loss facilitates progression of colorectal carcinoma.

35 Significance: These data show that the lncRNA SATB2-AS136 mediates epigenetic regulation of SATB2 and Snail expression37 to suppress colorectal cancer progression.

38

39 Introduction40 Colorectal carcinoma is the third most common cancer, with41 approximately 1.3 million new cancer cases and 690,000 mor-42 talities worldwide each year (1). Although the initial events in43 colorectal carcinoma have been relatively well-studied, and treat-44 ments for early-stage disease have significantly improved over the45 past decades, themechanisms ofmetastasis and relapse, which are46 the main causes of death, remain poorly characterized (2). Thus,47 gaining a better understanding of tumorigenesis and developing48 new diagnosis and treatment strategies for colorectal carcinoma is49 still urgently needed to improve colorectal carcinoma clinical50 outcome.51 Long noncoding RNAs (lncRNA), important new members of52 the ncRNA family, are functionally defined as transcripts >200

54nucleotides in length with no protein-coding potential. LncRNAs55have crucial functional importance in various biological process-56es, ranging from epigenetic gene regulation and transcriptional57control to posttranscriptional regulation (3, 4). Increasing evi-58dence has also established that the deregulation of lncRNAs59expressionmay be important in cancer biology, typically resulting60in the aberrant expression of gene products that contribute to the61progression of human tumors (5–8). For example, LncRNA-ATB,62which activated by the TGFb, competitively binds miR-200s to63upregulate ZEB1 and ZEB2 during the epithelial-to-mesenchymal64transition (EMT) and promote the invasion-metastasis cascade in65hepatocellular carcinoma (5). Thus, lncRNAs have been66highlighted as new players in cancer progression by functioning67as tumor suppressors, oncogenes or both, depending on the68circumstance. However, there are only preliminary studies on the69function of lncRNAs in colorectal carcinoma, and the overall70biological mechanisms and clinical significance of lncRNAs in71colorectal carcinoma remain largely unknown.72We speculated that there are still a large number of previously73unexplored lncRNAalterations in colorectal carcinoma, especially74colorectal carcinoma metastasis-related lncRNAs. We performed75lncRNA differential expression profiling between colorectal car-76cinoma tissues without metastasis and those with metastasis. We77found that lncRNA antisense transcript of SATB2 (SATB2-AS1)78was significantly down-regulated in colorectal carcinoma79tissues withmetastasis, which was associated with advanced stage80and poor prognosis of patients with colorectal carcinoma. The81effects of lncRNA SATB2-AS1 on colorectal carcinoma cell growth82and metastasis were assessed by gain- and loss-of-function

1Department of Pathology, Nanfang Hospital, Southern Medical University,Guangzhou, China.Q3 2Department of Pathology, School of Basic Medical Sciences,Southern Medical University, Guangzhou, China.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Y.-Q. Wang and D.-M. Jiang contributed equally to this article.

Corresponding Authors: Shuang Wang, Nanfang Hospital, #1838 GuangzhouNorth Road, Guangzhou, Guangdong 510515, China. Phone: 0086-020-62789364; Fax: 0086-020-61642148; E-mail: shuangw@126.com; andYan-Qing Ding, dyqgz@126.comQ4

doi: 10.1158/0008-5472.CAN-18-2900

�2019 American Association for Cancer Research.

AU

CancerResearch

www.aacrjournals.org 1

85 experiments in vitro and in vivo. We also unveiled themechanisms86 underlying the inhibitory effects of SATB2-AS1 on colorectal87 carcinoma progression via activating SATB2 transcription and88 then inhibiting EMT. This study provides a more detailed under-89 standing of SATB2-AS1 and its functions as a novel therapeutic90 target in colorectal carcinoma pathogenesis.

91 Materials and Methods92 Ethics statement93 The use of tissues for this study has been approved by the94 ethics committee of Nanfang Hospital, Southern Medical95 University (Guangzhou, China). All of the patients signed an96 informed consent, and the investigators obtained the written97 informed consent, before the use of these clinical materials98 for research purposes. This study was performed in strict99 accordance with the recommendations in the Guide for the100 Care and Use of Laboratory Animals of the National Institutes101 of Health. The protocol was approved by the Committee on102 the Ethics of Animal Experiments of Southern Medical103 University.

104 Microarray and computational analysis105 Three colorectal carcinoma tissues with metastasis and three106 colorectal carcinoma tissues without metastasis were used for107 the lncRNA expression profiling study. Briefly, total RNA was108 extracted from tissues and used to synthesize double-stranded109 cDNA. cDNA was labeled and hybridized to the LncRNA110 Expression Microarray (Arraystar). After hybridization and111 washing, slides were scanned with the Axon GenePix 4000B112 microarray scanner (Molecular Devices). The data were113 extracted using the NimbleScan software (Roche NimbleGen,114 Inc.). A P value was calculated using the t test. The threshold115 for differentially expressed genes was set as a fold change �2.0116 and a P value <0.05. The microarray data in this article117 have been deposited in NCBI Gene Expression Omnibus118 (GEO) and are accessible through (GEO) Series accession119 number GSE109910.

120 Tissue specimens and cell lines121 Fresh and formalin-fixed, paraffin-embedded colorectal tumor122 tissue samples were obtained from patients with a diagnosis of123 primary colorectal carcinoma and thenunderwent elective surgery124 in Nanfang Hospital, Southern Medical University. A total of 40125 cases of fresh colorectal carcinoma and paired nontumormucosal126 tissues were freshly frozen in liquid nitrogen and stored at�80�C127 until further use. And 176 cases of archived-colorectal carcinoma128 tissue samples were collected and used to investigate the expres-129 sion of SATB2-AS1 using ISH. Complete follow-up, ranging130 from 1 to 118 months, was available for the cohort of 176131 patients, and the median survival was 56 months. No patient132 received any pre-operative chemotherapy and radiotherapy.133 The human colorectal carcinoma cell lines DLD1, HCT116,134 SW480, SW620, LoVo, LS174t, and HT29 were obtained from the135 ATCC. A subclone named M5 with enhanced metastatic abilities136 in liver was isolated by in vivo selection of SW480 cells in our137 previous studies (9, 10). NCM460, a normal colon mucosa138 epithelial cell line (11), was obtained from the cell bank at the139 Chinese Academy of Sciences (Shanghai, China). Details of cell140 culture are summarized in the Supplementary Materials and141 Methods.

143RNA isolation and qRT-PCR144Total RNA was extracted using TRIzol Reagent (Ambion by life145Technologies). cDNA was synthesized using the Prime-Script RT146Reagent Kit (Promega). Two-step qRT-PCR was performed as147described previously (10). The results were normalized to the148expression of b-actin. The assay was performed in triplicate149for each case to allow for the assessment of technical variability.150The primer sequences used for PCR are listed in Supplementary151Table S1.

152ISH and evaluation of staining of SATB2-AS1153ISH was performed according to the manufacturer's protocol154(Exiqon). The ISH stained tissue sections were reviewed and155scored separately by two blinded pathologists. Staining for156SATB2-AS1 was assessed using a relatively simple, reproducible157scoring method (10). Additional details are described in the158Supplementary Materials and Methods.

159Construction of cell lines with stably overexpressed SATB2-AS1160The full-length human SATB2-AS1 with 3316-bp DNA frag-161ment amplified and cloned into GV303 lentiviral vector (Gene-162chem). Virus particles were harvested 48 hours after GV303-163SATB2-AS1 transfection cells using lipofectamine 2000 reagent164(ThermoFisher Scientific). Colorectal carcinoma cells were165infected with the recombinant lentivirus-transducing units plus1668 mg/mL Polybrene (Sigma) and then subjected to fluorescence-167activated cell sorting (FACS) analysis for GFP expression to gain168colorectal carcinoma cells with stable overexpression of SATB2-169AS1. The empty lentiviral vector GV303 was used as the control.

170Oligonucleotide transfection171The siRNAs SATB2-AS1, siRNA p300, siRNA SATB2, siRNA172Snail, and negative control siRNA (silencer negative control173siRNA) were synthesized by GenePharma (Shanghai, China).174Oligonucleotides transfection was performedwith Lipofectamine1752000 following the manufacturer's protocol. Target sequences for176siRNAs were shown in Supplementary Table S2.

177Functional assays in vitro178Cell proliferation, colony formation, wound-healing, and inva-179sion assays in vitro were performed according to standard proto-180cols as previously described (12, 13). The details are described in181the Supplementary Materials and Methods. All experiments were182performed in triplicate.

183In vivo tumorigenic and metastasis assays184Four-week-old female athyic Balb/C-nu/nu nude mice were185obtained from the Laboratory Animal Centre of SouthernMedical186University andmaintained in laminarflow cabinets under specific187pathogen-free conditions. For in vivo tumorigenicity, SATB2-AS1188overexpressing LoVo cells and empty vector stably transfected189cells were trypsinized, counted, and resuspended in sterile PBS. A190total of 5 � 106 SATB2-AS1 overexpressing LoVo cells or control191cells were subcutaneously injected into the right and left bilateral192upper limbs ofmice, respectively. The tumor volumes and overall193health of the mice were monitored. The size of the tumor was194determined by caliper measurements. Tumor volume was calcu-195latedwith the following the formula: 0.5� length�wildth2. Each196experimental group contained six mice.197For developing the in vivometastatic model, mice were injected198intravenously via the lateral tail vein with 2 � 106 cells. After 4

Wang et al.

Cancer Res; 2019 Cancer Research2

201 weeks of monitoring, the mice were sacrificed by cervical dislo-202 cation. The lungs were removed from the adjacent organs by203 dissection, and fixed using 10% neutral-buffered formalin. Sub-204 sequently, the consecutive tissue sections were obtained and205 stained with hematoxylin–eosin (H&E) to observe the metastatic206 nodules in the lungs under the microscope.

207 RNA immunoprecipitation208 RNA immunoprecipitation (RIP) assays were performed using209 the Magna RIP RNA-Binding Protein Immunoprecipitation Kit210 (Millipore) as described previously (7). Briefly, cells were cross-211 linked with 1% (w/v) formaldehyde and suspended in lysis212 buffer containing a protease inhibitor cocktail and an RNase213 inhibitor. Magnetic beads were preincubated with an anti-rabbit214 IgG or anti-rabbit p300 antibody (ab14984; Abcam) for 30min-215 utes at room temperature, and lysates were then immunopreci-216 pitated with the beads at 4�C overnight. RNA was purified from217 theRNA–protein complexes that bound to thebeads and thenwas218 analyzed by real-time RT-PCR.

219 Chromatin immunoprecipitation220 Chromatin Immunoprecipitation (ChIP) was performed using221 the EZ ChIP Chromatin immunoprecipitation Kit (Millipore),222 according to its manual. Briefly, cross-linked chromatin was223 sonicated into 200- to 1000-bp fragments. The chromatin was224 immunoprecipitated using anti-p300, anti-SATB2, or anti-225 HDAC1 antibody. Normal mouse immunoglobulin G (IgG) was226 used as a negative control. Real-time PCR was conducted using227 SYBR Green Mix (Takara). Primer sequences are listed in Supple-228 mentary Table S1.

229 RNA pull down assay230 RNA pull-down assay was performed as described previous-231 ly (14, 15). Briefly, Biotin-labeled SATB2-AS1 and its antisense232 RNAs were in vitro transcribed with the Biotin RNA Labeling Mix233 (Roche Diagnostics) and T7 RNA polymerase (Ambion by life234 Technologies). The cell lysates were freshly prepared using235 Whole Cell Lysis Assay (KGP2100; Keygen BioTECH). The Strep-236 tavidin Magnetic Beads (Life Technologies) were first prepared237 according to manufacturer's instructions and then immediately238 subjected to labeled RNA (50 pmol) capture in RNA capture239 buffer [20mmol/LQ6 Tris-HCl (pH7.5), 1MNaCl, 1mmol/L EDTA]240 for 30 minutes at room temperature with agitation. The RNA-241 captured beads were washed once with 20 mmol/L Tris (pH 7.5)242 and incubated with 1mg cell lysates diluted in 1X Protein-RNA243 Binding Buffer [20 mmol/L Tris (pH 7.5), 50 mmol/L NaCl,244 2mmol/L MgCl2, 0.1% Tween-20 Detergent] supplemented with245 50% glycerol for 2 hours at 4�C with rotation. The RNA-binding246 protein complexes were washed sequentially with Wash buffer247 [20 mmol/L Tris (pH 7.5), 10 mmol/L NaCl, 0.1% Tween-20248 Detergent] for three times, and eluted by Elution buffer. The249 eluted protein complexes were heated for 10 minutes at 100�C,250 separated by SDS-PAGE, and then analyzed by Western blotting251 assay. RNA inputwas detected using streptavidin-HRPby dot-blot252 assay.

253 Statistical analysis254 All the statistical analyses were performed using the SPSS255 version 16.0 software (SPSS). Differences between groups were256 identified using a two-tailed Student t test. Associations between257 SATB2-AS1 expression and clinicopathologic characteristics

259were determined by the x2 test. Survival curves were plotted260by the Kaplan–Meier method and compared by the log-rank test.261The significance of various variables for survival was analyzed by262the Cox proportional hazards model for multivariate analyses. A263probability value of 0.05 or less was considered to be significant.

264Results265Microarray analysis identifies deregulated lncRNAs related to266colorectal carcinoma metastasis267To identify novel functional lncRNAs in colorectal carcinoma268metastasis, we performed lncRNAmicroarray analysis to compare269lncRNA expression levels between colorectal carcinoma tissues270with and without metastasis. We found 140 upregulated and 168271downregulated lncRNAs in colorectal carcinoma tissues with272metastasis compared with those without metastasis (Fig. 1A).273Among them, we paid close attention to the top 20 lncRNAs in274each group. Of interest, the expression of lncRNA SATB2-AS1275(AK056625) was downregulated in the colorectal carcinoma276tissues with metastasis. SATB2-AS1 is an antisense cognate gene277of SATB2, a colorectal carcinoma metastasis suppressor gene278demonstrated in our previous studies (10, 16, 17). There is279evidence that antisense transcript lncRNAs can directly or indi-280rectly regulate the expression of the sense genes. Therefore, we281focused on SATB2-AS1 and investigated its clinical significance282and biological function in colorectal carcinoma.

283LncRNA SATB2-AS1 expression was significantly284downregulated in colorectal carcinoma tissues and cell lines285To validate the microarray analysis findings and investigate the286roles of SATB2-AS1 in colorectal carcinoma, we first measured287SATB2-AS1 expression in 40 colorectal carcinoma tissues and their288pair-matched noncancerous mucosa tissues. We found that289SATB2-AS1 transcript was expressed at lower levels in colorectal290carcinoma tissues compared with nontumor tissues of the same291donor (P ¼ 0.005, Fig. 1B). Significantly, downregulation of292SATB2-AS1 in tumor samples was associated with lymph-node293metastasis (P ¼ 0.013, Fig. 1B). A similar trend was observed in294unpaired colorectal carcinoma and adjacent normal samples from295The Cancer Genome Atlas (TCGA) cohort (P < 0.001, Fig. 1C).296Moreover, we also measured SATB2-AS1 levels in a panel of297colorectal carcinoma cell lines and a human colon mucosa298epithelial cell line (NCM460). Compared with NCM460,299SATB2-AS1 expression levels were significantly decreased in all300eight colorectal carcinoma cell lines (P < 0.001, Fig. 1D).

301Downregulation of SATB2-AS1 is associated with advanced302stage and poor prognosis of patients with colorectal carcinoma303To further explore the clinicopathologic and prognostic304significance of SATB2-AS1 expression, we investigated SATB2-AS1305expression in an independent panel of 176 primary colorectal306carcinoma tissues with clinical follow-up information using ISH.307SATB2-AS1-specific staining was localized in the nucleus of308benign and tumor epithelial cells (Fig. 1E). According to the309reclassification as described above, we divided the 176 patients310with colorectal carcinoma into a high SATB2-AS1 expression311group (n ¼ 72) and a low expression group (n ¼ 104). The low312SATB2-AS1 expression group showed an advanced T stage (P ¼3130.048), lymph node metastasis (P ¼ 0.004), and distant metas-314tasis (P ¼ 0.025) compared with the high SATB2-AS1 expression315group (Supplementary Table S3).

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 3

318 In addition, patients with low SATB2-AS1 expression had a319 significantly poorer prognosis than those with high SATB2-AS1320 expression (P ¼ 0.010, Fig. 1F). We also validated the prognostic321 significance of SATB2-AS1 in the TCGA cohort (n ¼ 525). As322 shown in Fig. 1G, Kaplan–Meier curve showed that low SATB2-323 AS1 expression was correlated with poor survival in patients with324 colorectal carcinoma (P¼ 0.035). Logistic multivariate regression325 revealed that low levels of SATB2-AS1 were independently asso-326 ciated with colorectal carcinoma overall survival [P ¼ 0.002;327 hazard ratio (HR), 0.537; 95% confidence interval (CI), 0.361–328 0.801). Taken together, we identified SATB2-AS1 expression level329 as an independent prognostic factor of disease outcome in330 patients with colorectal carcinoma.

332SATB2-AS1 overexpression represses colorectal carcinoma cell333proliferation334To elucidate causal roles of SATB2-AS1 in colorectal carcinoma,335we applied lentivirus delivery of SATB2-AS1 to ectopically336overexpress SATB2-AS1 in two colorectal carcinoma cell lines,337M5 and LoVo cells. Increased expression of SATB2-AS1 after338infection of lentivirus was confirmed in the two cell lines by339real-time RT-PCR (P < 0.05, Supplementary Fig. S1). The CCK8340assays revealed a significantly slower proliferation rate in SATB2-341AS1-overexpressing M5 and LoVo cells when compared with342control cells (P < 0.001, Fig. 2A). Meanwhile, SATB2-AS1 over-343expression suppressed colony formation in M5 (P < 0.001) and`344LoVo cells (P ¼ 0.023, Fig. 2B).

Figure 1.

LncRNA SATB2-AS1 is downregulated in colorectal carcinoma, especially in colorectal carcinomawith metastasis and this lncRNA could be an independentprognostic factor for the prediction of the overall survival of patients with colorectal carcinoma. A, Hierarchical clustering analysis of lncRNAs that weredifferentially expressed in colorectal carcinoma tissues with and without metastasis. C denotes colorectal carcinoma tissues without metastases and CM denotescolorectal carcinoma tissues with lymph node metastases. B, The level of SATB2-AS1 in paired colorectal carcinoma and adjacent noncancerous tissues. nmCRCdenotes colorectal carcinoma tissues without metastases and mCRC denotes colorectal carcinoma tissues with lymph nodemetastases. C, The level ofSATB2-AS1 in unpaired colorectal carcinoma and noncancerous tissues samples from TCGA cohort. D, The level of SATB2-AS1 in colorectal carcinoma and colonmucosa epithelial (NCM460) cell lines. E, Expression analysis of SATB2-AS1 in normal colorectal mucosa and colorectal carcinoma tissues by ISH. (a) Positiveexpression of SATB2-AS1 in normal colorectal mucosa. (b–d) Negative, medium, and high expression of SATB2-AS1 in colorectal carcinoma tissue, respectively.Scale bars are shown in the right left corner of each picture. F and G, Kaplan–Meier analysis of overall survival in all patients with colorectal carcinoma accordingto SATB2-AS1 expression according to our data (F) and TCGA cohort (G).Q7

Wang et al.

Cancer Res; 2019 Cancer Research4

Figure 2.

SATB2-AS1 overexpression inhibits colorectal carcinoma cell growth andmetastasis in vitro and in vivo.A, SATB2-AS1 overexpression suppressed cellproliferation in colorectal carcinoma cell lines as determined by CCK-8 assay. B, SATB2-AS1 overexpression inhibited colony formation in colorectal carcinomacells. Representative images (left) and quantitative analyses (right) are shown. C, SATB2-AS1 overexpression induced cell-cycle arrest in G1 phase in colorectalcarcinoma cells. D, SATB2-AS1 overexpression inhibited colorectal carcinoma cell invasion in Matrigel invasion assay. E, SATB2-AS1 overexpression suppressedcell migration in the wound-healing assay in M5 and LoVo cells. The experiments were performed at least in triplicate, and the data are expressed as themean� SD. F, SATB2-AS1 overexpression inhibited subcutaneous tumor formation in nudemice. LoVo cells with ectopic overexpression of SATB2-AS1 andcontrol cells were inoculated into nudemice (n¼ 6 per group). These graphs show the tumor xenografts 3 weeks after ectopic-subcutaneous implantation innude mice. The effect of SATB2-AS1 on colorectal carcinoma tumor growth was evaluated based on tumor volume in the two groups. G, Representativephotographs of hematoxylin and eosin (H&E) and IHC staining for Ki-67 and SATB2 antibody of primary cancer tissues. H, Representative pictures of lungmetastasis by H&E staining in nudemice 4 weeks after tail vein injection with SATB2-AS1 overexpression LoVo cells or control cells (n¼ 8 per group). I, Statisticalcomparisons of lung metastasis and pulmonary tumor colonies in the two groups of mice after tail vein injection. � , P < 0.05; �� , P < 0.01; ��� , P < 0.001.

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 5

347 We next probed the potential mechanisms underlying the348 growth-inhibitory effects of SATB2-AS1 overexpression. Flow349 cytometry cell-cycle assays in these cell lines demonstrated that350 SATB2-AS1 overexpression induced G1 phase cell-cycle arrest in351 M5 (P ¼ 0.035) and LoVo cells (P < 0.001, Fig. 2C). However,352 the proportion of apoptotic cells remained similar between353 SATB2-AS1-overexpressing cells and controls cells (P < 0.05,354 Supplementary Fig. S2). These data suggest that the SATB2-355 AS1-mediated decrease in colorectal carcinoma cell proliferation356 is modulated by the G1–S checkpoint, rather than by apoptosis.

357 SATB2-AS1 overexpression inhibits colorectal carcinoma cell358 migration and invasion359 We also examined the effects of SATB2-AS1 overexpression on360 colorectal carcinoma cell migration and invasion. Cell invasion361 analysis demonstrated that enforced expression of SATB2-AS1362 in two different colorectal carcinoma cell lines significantly363 inhibited cancer cell invasion through Matrigel, a basement-364 membrane-like extracellular matrix (P < 0.001, Fig. 2D). The365 wound-healing assay also illustrated that the exogenous expres-366 sion of SATB2-AS1 in colorectal carcinoma cells caused a signif-367 icant decrease in cell migration (M5 cells, P ¼ 0.013 and LoVo368 cells, P < 0.001, Fig. 2E). These findings indicated that SATB2-AS1369 was sufficient to repress both colorectal carcinoma cell invasion370 and migration in vitro.

371 SATB2-AS1 overexpression suppresses tumorigenicity and372 metastasis of colorectal carcinoma cells in vivo373 In light of our in vitro findings, we tested the effects of SATB2-374 AS1 overexpression in vivo. LoVo cells stably overexpressing375 SATB2-AS1 and control cells were subcutaneously inoculated into376 nude mice (n ¼ 6 per group). Mice injected with cells over-377 expressing of SATB2-AS1 showed a measurably suppressed effect378 on tumor growth compared with those injected with control cells379 (Fig. 2F). SATB2-AS1 overexpression significantly inhibited cell380 proliferation in LoVo cells as determined by Ki-67 staining381 (Fig. 2G). We further explored the role of SATB2-AS1 in lung382 colonization by inoculating cells directly into the tail veins of383 nude mice (n¼ 8 per group). The results showed that SATB2-AS1384 overexpression decreased the number of definite pulmonary385 colonization (oneof eightmice) comparedwith the control group386 (five of eight mice; P ¼ 0.039, Fig. 2H and 2I). Moreover,387 compared with mice injected with control cells, the number of388 pulmonary tumor colonies in the SATB2-AS1-overexpressing cell-389 injected mice was significantly decreased (P ¼ 0.041, Fig. 2I).390 Consistent with the in vitro results, these data also indicated an391 important inhibitory role for SATB2-AS1 in tumor growth and392 metastasis in vivo.

393 siRNA-mediated knockdown of SATB2-AS1 expression394 promotes colorectal carcinoma cell proliferation, invasion, and395 migration396 To further confirm the effects of SATB2-AS1 on suppressing the397 malignant phenotypes in colorectal carcinoma cells, we also398 knocked down SATB2-AS1 expression by siRNA transfection in399 SW480 and DLD1 cells, which have relatively high SATB2-AS1400 expression. SW480 cells with depleted SATB2-AS1 expression401 exhibited enhanced cell growth by more than 40% at day 4402 (P ¼ 0.002) and DLD1 cells showed 20% enhanced cell growth403 at day 6 (P ¼ 0.049), compared with negative control scrambled404 siRNA transfection (Supplementary Fig. S3A). Furthermore,

406colony formation assays revealed that SATB2-AS1 siRNA-treated407cells exhibited significantly increased cell growth compared408with the negative control siRNA (SW480, P ¼ 0.026 and DLD1,409P ¼ 0.001, Supplementary Fig. S3B). SATB2-AS1 knockdown410reduced the G1 phase and increased the S-phase cell population411in SW480 (P ¼ 0.013) and in DLD1 cells (P < 0.001, Supple-412mentary Fig. S3C). Moreover, we also found that depletion413of SATB2-AS1 increased the invasion and migration potential of414colorectal carcinoma cells, compared with the control cells415(SW480, P < 0.001 and P ¼ 0.041 and DLD1, P < 0.001 and416P¼ 0.007, Supplementary Fig. S3D and S3E). Consistent with the417gain-of-function results, these data also suggest that SATB2-AS1 is418a suppressor in colorectal carcinoma cells.

419SATB2, the sense-cognate gene for SATB2-AS1, is a key420downstream target of SATB2-AS1421Given the close proximity of SATB2-AS1 to SATB2 (Fig. 3A), we422hypothesized that SATB2-AS1 could exert its biological effects via423SATB2 modulation. To examine whether SATB2-AS1 is co-424expressed with SATB2 in human colorectal carcinoma samples,425wemeasured SATB2mRNA levels in the same set of 40 colorectal426carcinoma tissues shown in Fig. 1B. Similar to SATB2-AS1, SATB2427mRNA was significantly downregulated in the majority of colo-428rectal carcinoma samples (P < 0.001, Supplementary Fig. S4),429and SATB2-AS1 and SATB2 RNA levels in these tissues were430positively correlated (R2 ¼ 0.1698, P ¼ 0.008, Fig. 3B left). As431shown in Fig. 3B (right), a significantly positive correlation432between SATB2-AS1 and SATB2 was found in the TCGA cohort433(n¼ 623, R2¼ 0.6222, P < 0.001). Furthermore, wemeasured the434levels of SATB2 in colorectal carcinoma cells with overexpressing435SATB2-AS1. SATB2-AS1 overexpression significantly increased436SATB2 mRNA levels in M5 and LoVo cells (M5 cells, P < 0.001437and LoVo cells, P ¼ 0.033, Fig. 3C, top left). When measured by438western blotting, the expression of SATB2proteinwas increased in439both M5 and LoVo cells with overexpressing of SATB2-AS1440(Fig. 3C, top right).Meanwhile, SATB2-AS1 knockdownmediated441by siRNA significantly reduced SATB2expression comparedwith a442negative control siRNA (Fig. 3C, bottom). These observations443confirmed the regulation of SATB2 expression by SATB2-AS1444in vitro at the mRNA and protein level. Furthermore, the445results from mouse models showed that the overexpression of446SATB2-AS1 could significantly increase SATB2 protein levels447in vivo (Fig. 2G, bottom). Of note, the depletion of SATB2 also448affected SATB2-AS1 expression levels to some extent (Supplemen-449tary Fig. S5). Taken together, these results suggest an important450role for SATB2-AS1 in modulating the expression of SATB2, the451sense-cognate gene for SATB2-AS1, and vice versa.

452SATB2-AS1 promotes SATB2 expression by recruiting p300 to453accelerate histone H3 acetylation454We propose that the overlapping parts of the SATB2-AS1 and455SATB2 transcripts could form an RNA duplex to further alter the456secondary or tertiary structure of SATB2, which increases the457stability of SATB2. As a result, we assessed the stability of the458SATB2 transcript by blocking new RNA synthesis with Dactino-459mycin D. We did not observe increased SATB2 mRNA stability in460colorectal carcinoma cells stably overexpressing SATB2-AS1 com-461pared with control cells (P > 0.05, Supplementary Fig. S6).462Nuclear and cytoplasmic total RNA fractions were prepared463from colorectal carcinoma cells. As shown in Fig. 4A, SATB2-AS1464was 377-fold enriched in the nuclear fraction relative to the

Wang et al.

Cancer Res; 2019 Cancer Research6

467 cytoplasm, suggesting that the SATB2-AS1 transcript was mainly468 located in the nucleus. A similar subcellular location for SATB2-469 AS1 in colorectal carcinoma cells was confirmed by fluorescence

471ISH (FISH, Fig. 4A, right). Because nuclear-enriched lncRNAs472are known to be functionally involved in epigenetic and tran-473scriptional regulation, we wondered whether the mechanism of

Figure 3.

SATB2, the sense-cognate gene for SATB2-AS1, is a key downstream target of SATB2-AS1.A, Genomic location of SATB2 and SATB2-AS1 from ENCODEcollection. Higher levels of epigenetic modification marks on histone 3 at lysines 4 (H3K4me3), 9 (H3K9Ac), and 27 (H3K27Ac) and several transcription factorbinding sites uniform peaks of p300 were observed within the SATB2 promoter region. B, The correlation between SATB2-AS1 transcript levels and SATB2mRNA levels in colorectal carcinoma tissues was measured according to our data (n¼ 40, left) and the TCGA cohort (n¼ 623, right). C, SATB2-AS1 regulated theexpression of SATB2 onmRNA and protein levels.

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 7

Figure 4.

SATB2-AS1 promotes the expression of SATB2 by recruiting p300 to accelerate histone H3 acetylation. A, Nuclear-enriched SATB2-AS1was determined by RNAfraction assays (left) and fluorescence in situ hybridization (right). B, RIP experiments were performed in LoVo cells using a p300 or nonspecific IgG antibody todetermine the amount of SATB2-AS1 RNA associated with p300 or IgG relative to the input control. C, RNA pull-down assays were used to identify proteinsassociated with SATB2-AS1. Biotinylated SATB2-AS1 and antisense RNAwere incubated with cell extracts and the associated proteins were resolved bySDS-PAGE.Western blotting was performed to analysis the specific interaction between p300 and SATB2-AS1. Dot-blot of RNA-protein binding samplesindicates equal RNA transcript present in the assay.D, RNAs corresponding to different fragments of SATB2-AS1 or its antisense sequence (dotted line) werebiotinylated and incubated with LoVo cell extracts, targeted with streptavidin beads. p300 protein was detected byWestern blotting. E, SATB2-AS1overexpression significantly increased the protein level of acetylated H3K9 and H3K27, and SATB2, and did not affect the protein level of tri-methylated H3K4and p300. F and G, p300 could bind to SATB2 promoter. Schematic illustration of SATB2-AS1 and SATB2 structures was shown (F upper). Arrows show thedirection of transcription. SATB2-AS1 exons are depicted as red bars and marked as E1–E3. SATB2 exons are depicted as blue bars. There are 164 nucleotideoverlapping and complementary regions between the second exon of SATB2-AS1 and SATB2 exon1. The numbered sites denote the promoter fragments ofSATB2 gene. The ability of p300 to bind to SATB2 promoter regions was assessed by ChIP (F) and was increased as a result of SATB2-AS1 overexpression (G).H, SATB2 protein was upregulated by the histone deacetylase inhibitor TSA in colorectal carcinoma cells, whose effect was equal to that of SATB2-AS1overexpression. The induction of SATB2 expression by SATB2-AS1 could be reversed by the acetyltransferase inhibitor C646. I, Depletion of p300mediated bysiRNA significantly increased H3K9ac and H3K27ac acetylation and SATB2 protein levels in colorectal carcinoma cell lines with overexpressing SATB2-AS1. J,H3K9ac, H3K27ac, and SATB2 levels were upregulated with the increasing p300 recruited by SATB2-AS1 overexpression, whereas H3K9ac, H3K27ac, and SATB2were downregulated when SATB2-AS1was lowly expressed in tissues from the same donor gland in human colorectal carcinoma.

Wang et al.

Cancer Res; 2019 Cancer Research8

476 SATB2-AS1 promotes SATB2 expression through modulation of477 transcription factor recruitment and chromatin modification. We478 performed a computational screen. ENCODE data in the UCSC479 Genome Browser showed that higher levels of epigenetic mod-480 ification marks on histon 3 at lysines 4 (H3K4me3), 9 (H3K9ac),481 and 27 (H3K27ac) and several transcription factor binding sites482 uniform peaks of p300, a histone acetyltransferase, within the483 SATB2 promoter region in K562 cells (Fig. 3A). Then, we per-484 formed RIP with an antibody against p300 using colorectal485 carcinoma cells extracts, and observed significant enrichment of486 SATB2-AS1 with the p300 antibody compared with the nonspe-487 cific IgG control antibody (P¼ 0.005, Fig. 4B). The previous study488 confirmed that lncRNA ZEB1-AS1 recruited p300 to ZEB1 pro-489 moter, upregulating histone markers at the ZEB1 promoter (18).490 As expected, p300was able to interactwithZEB1-AS1 in colorectal491 carcinoma cells extracts with p300 antibody. Meanwhile, the492 p300 binding to SATB2-AS1 was specific, because of failing to493 detect the presence of PACER, a lncRNA reported that could not494 interact with p300 (19), in p300RIP samples (Supplementary Fig.495 S7A–S7B). Furthermore, RNA pull-down was performed to val-496 idate the association between SATB2-AS1 and p300 in colorectal497 carcinoma cells, which confirmed the physical association498 between SATB2-AS1 and p300 in vitro (Fig. 4C). Deletion-map-499 ping analyses identified the 361�720 nt region at the 50 end of500 SATB2-AS1 that is required for its associationwith p300 (Fig. 4D).501 However, SATB2-AS1 could not affect p300 mRNA and protein502 expression levels (Fig. 4E, top; Supplementary Fig. S8). Subse-503 quently, we performed ChIP using an anti-p300 antibody to504 determine whether p300 was bound to the SATB2 promoter505 region and SATB2-AS1 overexpression affects its binding capa-506 bility. We observed a significant enrichment of SATB2 promoter507 fragmentswith the p300 antibody (Fig. 4F).Moreover, a 23.3- and508 3992.1-fold increase in SATB2 promoterDNAbound to p300was509 identified in cells with overexpressing SATB2-AS1 compared with510 control cells expressing endogenous levels of SATB2-AS1 (P <511 0.05, Fig. 4G). Taken together, these results demonstrate that512 SATB2-AS1 served as a scaffold to recruit p300 to the SATB2513 promoter.514 We next addressed whether SATB2-AS1 overexpression, which515 increases p300 occupancy, also affected the levels of H3K9 and516 H3K27 acetylation and H3K4 tri-methylation (H3K4me3) in the517 SATB2 promoter region. As shown in Fig. 4E, there was a signif-518 icant increase in the protein level of acetylated H3K9 and H3K27,519 and SATB2 as a result of SATB2-AS1 overexpression. However, no520 significant difference was observed in the protein level of521 H3K4me3. We also treated the colorectal carcinoma cells with522 the histone deacetylase inhibitor trichostatin A (TSA) and acet-523 yltransferase inhibitor C646. Interestingly, SATB2 was upregu-524 lated by TSA in both LoVo and M5 cells, and that the effect was525 equal to that of SATB2-AS1 overexpression. Meanwhile, the526 induction of SATB2 expression by SATB2-AS1 could be reversed527 by C646, as shown in Fig. 4H.528 Accordingly, we addressed whether the enhancement of529 H3K9ac and H3K27ac by SATB2-AS1 was mediated by p300. In530 colorectal carcinoma cells with overexpressing SATB2-AS1 trans-531 fected with specific p300 siRNAs, the protein levels of H3K9ac,532 H3K27ac, and SATB2 were significantly decreased compared to533 the cells transfected with negative control siRNA (Fig. 4I). Con-534 sistent with the in vitro results, p300, H3K9ac, H3K27ac, and535 SATB2 levels were upregulated with SATB2-AS1 overexpression in536 human colorectal carcinoma tissues. However, with SATB2-AS1

538low-expression, p300, H3K9ac, H3K27ac, and SATB2 levels were539downregulated from the same donor gland (Fig. 4J). These results540support the hypothesis that the transcript activation of SATB2 is541dependent on the level of SATB2-AS1 recruiting p300, which542induces H3K9 and H3K27 acetylation in the SATB2 promoter543region.

544SATB2-AS1 requires p300/SATB2 to suppress colorectal545carcinoma growth, invasion, and migration546To test the contribution of p300 and SATB2 on the important547role of SATB2-AS1, we knocked down the expression level of p300548or SATB2 in vector- or SATB2-AS1-expressing colorectal carcino-549ma cells with siRNAs targeting p300 or SATB2, respectively.550SATB2-AS1 overexpression significantly reduced LoVo cell prolif-551eration, cell-cycle progression, invasion, andmigration.However,552this potential was completely abolished to the levels similar to the553control cells when knockdown of p300 mediated by siRNA554(Fig. 5A–E). These results suggest that p300 is specifically required555for SATB2-AS1 to affect LoVo cells behaviors. Moreover, we556observed that depletion of SATB2 impairs LoVo cell proliferation557efficiency and diminishes the suppressive effects of SATB2-AS1558overexpression in a similar manner (Fig. 5A–E). Similar results559were obtained using M5 cells (Supplementary Fig. S9A–S9E).560Taken together, these rescue effects obtained by knocking down561p300or SATB2 indicated that the ability of SATB2-AS1 to suppress562growth, invasion, and migration is in large part attributed to its563ability to recruit p300, which subsequently enhances H3K9 and564H3K27 acetylation levels in the SATB2 promoter and activates565SATB2 gene transcription.

566Suppression of EMT by SATB2-AS1 depends on SATB2-567mediated recruitment of HDAC1 to repress Snail transcription568Our previous studies showed that SATB2 was associated with569cell invasion and migration in colorectal carcinoma (10, 16, 17).570Gene set enrichment analysis (GSEA) revealed that lower expres-571sion of SATB2 was positively correlated with an enrichment of572metastasis signatures (Fig. 6A, GSE17536, GSE13067, and573GDS4718). Indeed, SATB2 downregulation, directly regulated by574miR-182, induced colorectal carcinoma cells EMT (16), which575indicated that SATB2 downregulation-induced EMT might be a576mechanism that SATB2-AS1 depletion facilitated colorectal car-577cinoma progression. Interestingly, both SATB2-AS1- and SATB2-578overexpression resulted in significantly decreased mRNA and579protein levels of Snail, a central regulator of EMT (Fig. 6B and C).580To further explore the underlying mechanism of how SATB2581regulates Snail expression, we searched Snail promoter sequence582with BLAST software and identified that three potential binding583regions with AT-rich sequence (42.5%, 29.3%, and 30.8%) were584predicted to be embedded in Snail gene promoter (Fig. 6D, top).585ChIP assays revealed that endogenous SATB2 protein was bound586to the most AT-rich region (P1, �1069 to �710 bp) of Snail587promoter (Fig. 6D).588Recruitment of histone deacetylase (HDAC) has been shown as589typical functional characteristics of MAR-binding proteins (20).590To investigate whether HDACs is involved in the regulation of591Snail, we immunoprecipitated protein complex from colorectal592carcinoma cells extracts using anti-SATB2 antibody with normal593mouse IgG as the negative control and detected HDAC1 and594HDAC2 protein in the precipitates, and found that SATB2 co-595precipitated with HDAC1 but not with HDAC2. The interaction596between SATB2 and HDAC1 was further confirmed by detecting

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 9

599 SATB2 in the immunoprecipitate of HDAC1 in DLD1 cells600 (Fig. 6E).601 As histone deacetylation is tightly coupled to HDAC-mediated602 gene repression, we therefore examined HDAC1 occupancy at the603 Snail promoter by ChIP assays. We observed a significant enrich-604 ment of Snail promoter (P1 and P2) DNA with anti-HDAC1605 antibody. The ability of HDAC1 binding to the Snail promoter606 (P1) was significantly increased in colorectal carcinoma cells607 overexpressing SATB2 compared with control cells (Fig. 6F),608 which was consistent with the above results that the promoter609 region of Snail interacted with SATB2. Meanwhile, the SATB2610 downregulation significantly increased Snail promoter luciferase611 activity (Fig. 6G). These results indicated that SATB2 recruit612 HDAC1 to Snail promoter to repress transcription by reducing613 levels of acetylated histone. In addition, we also observed that614 p300 occupied at the Snail promoter through ChIP assays. How-615 ever, SATB2-AS1-overexpression could not significantly increase616 this occupancy ability (Supplementary Fig. S10), suggesting that617 p300 recruitment by SATB2-AS1 cannot enhance the acetylation618 levels of Snail promoter. Interestingly, the depletion of Snail619 affected SATB2-AS1 expression levels to some extent (Supplemen-620 tary Fig. S11).621 To examine the contribution of Snail to the phenotype caused622 by SATB2-AS1 overexpression, we performed rescue experiments.623 We increased Snail level in SATB2-AS1-overexpressing and624 control cells by transfecting Snail overexpression plasmid. Snail625 overexpression significantly repaired the inhibiting effects caused626 by overexpression of SATB2-AS1 on colorectal carcinoma cell627 proliferation, invasion, and migration to the levels similar to the

629control cells (Figure 7A–E). These results suggest that Snail is630specifically required for SATB2-AS1 to affect LoVo cells behaviors.631Moreover, we also found that Snail overexpression rescued the632suppressive effects of SATB2 overexpression in a similar manner633(Fig. 7A–E).634Snail has been proposed as a central regulator of EMT. There-635fore, we next further investigated the expression changes of classic636EMTmarkers. As expected, a significant gain of epithelial markers637(E-cadherin and ZO-1) and loss of the mesenchymal marker638vimentin was observed both in SATB2-AS1- and SATB2-over-639expression colorectal carcinoma cells. However, these changes640were completely attenuated to the levels similar to the control641cells by HDAC1 depletion (Fig. 7F). Taken together, these results642indicate that suppression of colorectal carcinoma progression by643SATB2-AS1 depends on SATB2-mediated recruiting HDAC1 to644repress Snail transcription and EMT (Fig. 7G).

645Discussion646In this study, we uncovered a long noncoding antisense tran-647script for SATB2, called SATB2-AS1, which functions as a regulator648of SATB2 gene expression. We present data showing that the649SATB2-AS1 expression levels were downregulated in colorectal650carcinoma tissues and that the downregulation of SATB2-AS1651expression was intimately linked to the TNM classification and652poor outcome of patients with colorectal carcinoma. We also653validated that SATB2-AS1 played an important suppressed role654during colorectal carcinoma tumorigenesis and progression.655Additionally, we demonstrated that SATB2-AS1 recruited p300

Figure 5.

SATB2-AS1 requires p300/SATB2 to suppress colorectal carcinoma cell growth, invasion, and migration. SATB2-AS1 overexpression significantly reduced thecolorectal carcinoma cell proliferation (A), colony formation (B), cell-cycle progression (C), migration (D), and invasion (E). The potential effects of SATB2-AS1 oncolorectal carcinoma cells were completely abolished similar to the control cells by the p300-mediated knockdown by siRNA. The depletion of SATB2 impairscolorectal carcinoma tumor cell proliferation efficiency and diminishes the suppression effect of SATB2-AS1 overexpression in a similar manner.

Wang et al.

Cancer Res; 2019 Cancer Research10

Figure 6.

SATB2-AS1 repressed Snail transcription depending on SATB2-mediated recruitment of HDAC1. A,GSEA showed the enrichment of metastasis signatures incolorectal carcinoma cells with SATB2 down-regulation. B and C, Both SATB2-AS1 and SATB2 inhibited the expression of Snail on mRNA and protein levels. D,ChIP analysis of SATB2 occupancy on the Snail promoter. Immunoprecipitated DNA was compared with input and shown as percentage. Data are represented asmean� SD. E, Co-immunoprecipitation of endogenous HDAC1 with anti SATB2 antibodies (top) and endogenous SATB2 with anti HDAC1 antibodies (bottom) incolorectal carcinoma cells. F, ChIP analysis of HDAC1 occupancy on the Snail gene. Significantly increased recruitment of HDAC1 to Snail promoter was observedupon SATB2 overexpression. G,Dual luciferase reporter assay was performed in colorectal carcinoma cells to verify the effect of SATB2 on Snail promoteractivity.

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 11

Figure 7.

SATB2-AS1 requires SATB2/Snail to suppress colorectal carcinoma cell growth, invasion, migration, and EMT. SATB2-AS1 overexpression significantly reducedthe colorectal carcinoma cell proliferation (A), colony formation (B), cell-cycle progression (C), invasion (D), and migration (E). The potential effects ofSATB2-AS1 on colorectal carcinoma cells were completely abolished similar to the control cells by Snail overexpressing. Snail overexpression also rescued thesuppressive effects of SATB2 overexpression in a similar manner (A–E). F,Western blot analysis showed that epithelial markers were reduced and vimentin as amesencymal marker was enhanced both by SATB2-AS1 and SATB2-overexpression. The changes were completely attenuated to the levels similar to the controlcells by HDAC1 depletion in colorectal carcinoma cells. G, Schematic diagram showing the mechanism of action of SATB2-AS1 in colorectal carcinoma. SATB2-AS1was downregulated in colorectal carcinoma and consequently decreased p300 recruitment. In turn, the reduction in p300 recruitment suppressed theaccumulation of the active marks H3K27ac and H3K9ac and repressed SATB2 levels. Subsequently, the decreasing of HDAC1 recruited by SATB2 increasedactivating histone acetylation marks and promoted Snail expression, which induced EMT and colorectal carcinoma progression.

Wang et al.

Cancer Res; 2019 Cancer Research12

658 to the SATB2promoter and increasedH3K9ac andH3K27ac levels659 at the SATB2 promoter region, resulting inmarked transcriptional660 activation of SATB2 gene. Then, SATB2 inhibited EMT through661 recruiting HDAC1 to repress Snail transcription. Thus, our results662 reveal a novel mechanism of SATB2-AS1-SATB2-Snail regulatory663 axis involved in epigenetic regulation and the SATB2-AS1-SATB2-664 Snail pathway constitutes a previously unrecognized carcinogen-665 esis and progression regulator in colorectal carcinoma.666 Recent advances in genomics technology have led to the explo-667 sive discovery that mammalian genomes encode numerous668 lncRNAs (3), which advancing our comprehensive understanding669 of biological processes in human health and disease. It has been670 identified that thousands of lncRNAs that are differentially tran-671 scribedbetweennormal tissues and tumors (21, 22). Validationof672 these expression-deregulated lncRNAs may provide biomarkers673 for diagnosis, determining prognosis and as a therapeutic target.674 We applied high-throughput lncRNA expression profiling to675 detect colorectal carcinoma metastasis-related lncRNAs which676 might play critical roles in colorectal carcinoma progression677 processes. Among them, we focused on an antisense lncRNA,678 SATB2-AS1, and then demonstrated that SATB2-AS1 was down-679 regulated in colorectal carcinoma tissues and cell lines and that680 the low SATB2-AS1 expression levels were significantly associated681 with aggressive stages andpoor survival of patientswith colorectal682 cancer. These data indicate that the expression of SATB2-AS1may683 have considerable potential in predicting the prognosis of colo-684 rectal carcinoma.685 Accumulating data are revealing the extensive functional roles686 of lncRNAs in tumorigenesis. Gupta and colleagues reported that687 HOTAIR induced genome-wide retargeting of PRC2, leading to688 H3K27me3, which promoted breast cancer metastasis by silenc-689 ing multiple metastasis suppressor genes (8). Subsequent studies690 showed that HOTAIR deregulation is associated with patient691 progression in 26 human tumor types (23). LncRNA CCAL692 enhances proliferation, cell-cycle progression, invasion, and693 migration, which are in large part attributed to its ability to694 degrade AP-2a and subsequently activate the Wnt/b-catenin695 signaling pathway (24). Very recently, one study showed that696 elevated expression of SATB2-AS1 increases cell proliferation and697 growth in osteosarcoma (25). In this study, we comprehensively698 confirmed the effects of SATB2-AS1 on colorectal carcinoma699 carcinogenesis through gain- and loss-of-function experiments.700 We identified that SATB2-AS1 plays a key role in colorectal701 carcinoma carcinogenesis by acting as a tumor suppressor which702 inhibits colorectal carcinoma growth and metastasis in vitro and703 in vivo. The results also indicate that the enhanced expression of704 SATB2-AS1 by gene transfer can reverse the malignant phenotype705 of colorectal carcinoma, suggesting SATB2-AS1 as a potential706 therapeutic target for colorectal carcinoma.707 Many new functions have been ascribed to antisense lncRNAs708 from data generated over the last year. However, the mechanisms709 involved in mediating the progression of colorectal carcinoma by710 SATB2-AS1 remain unknown. The sense-cognate gene of SATB2-711 AS1, SATB2, one of matrix attachment region (MAR)-binding712 proteins, binds to AT-rich MARs of DNA and induces a local713 chromatin-loop remodeling, controllinggeneexpression(20,26).714 In our previous studies, we found that SATB2 is ubiquitously715 expressed in noncancerous tissues; in contrast, its expression is716 frequently low in tumors and tumor cell lines, leading to increased717 colorectal carcinoma cell proliferation and metastasis (10, 16,718 17).However, themechanisms involved in deregulating of SATB2

720expressionhave not been clearly defined. Remarkably, SATB2-AS1721abundance through ectopic overexpression or siRNA-mediated722knockdown specifically affects SATB2 mRNA and protein expres-723sion. Moreover, SATB2-AS1 and SATB2 RNA levels in colorectal724carcinoma tissues were significantly correlated. Thus, we specu-725lated that SATB2-AS1 could exert its biological effects via726SATB2 modulation. Previously, studies have revealed that the727antisense might form sense-antisense pairs to regulate epigenetic728silencing, transcription, mRNA stability, and gene translation on729the opposite strand (27, 28). For example, lncRNA BACE1-anti-730sense transcript (BACE1-AS) increases BACE1 mRNA stability by731generating additional Ab1-42 in Alzheimer's disease (27). How-732ever, SATB2-AS1 failed to significantly alter the SATB2 mRNA733degradation rate. Thus, SATB2-AS1 does not appear to affect the734stability of SATB2 by forming form sense-antisense gene pairs.735Apart from proteins, the functionality of lncRNAs begins with736their cellular localization. These so-called nuclear-enriched737lncRNAs are shown to be involved in key cellular processes738associated with gene expression including but not limited to739epigenetic regulation, chromosomal interaction, and transcrip-740tional regulation (29).Here, SATB2-AS1wasdetermined to enrich741in nucleus of CRC cells by ISH, FISH, and RNA fraction assays,742which is consistent with a potential role in epigenetic regulation.743Indeed, SATB2-AS1 could bind with p300 protein, thereby induc-744ing the accumulation of the active marks H3K9ac and H3K27ac,745which activate SATB2 transcription. Importantly, we also746observed that changes in SATB2-AS1 abundance specifically747affected SATB2 expression throughp300 recruitment in vivo. These748data suggest SATB2-AS1 epigenetically regulate the SATB2 pro-749moter at the histone level. Mutiple lncRNAs have previously been750described to regulate genes expression via epigenetic regulation,751such as the recently identified lncRNA ANRASSF1 (28) and752NEAT1 (30). Our results, along with recent studies, indicate that753lncRNAs have the potential to influence the epigenetic environ-754ment of genes in a location-specific manner. Furthermore, the755most important effect exerted by SATB2-AS1 on cell proliferation,756invasion, and migration is partially reversed by the siRNA-757mediated knockdown of p300 or SATB2 after transfection with758siRNAs. Thus, our results reveal a novel mechanism for epigenetic759regulation at SATB2 that involves the antisense lncRNA SATB2-760AS1 and demonstrates that the SATB2-AS1/SATB2 pathway con-761stitutes an important role in colorectal carcinoma progression.762EMT is an evolutionary conserved trans-differentiation process763proposed to play a key role in cancer metastasis, therapeutic764resistance and cancer stem cell-like features (31–33), which are765essential features for colorectal carcinoma progression (34–36).766During EMT, epithelial cells lose their epithelial adherence, lose767their polarity, attain front-to-real polarity, become spindle768shaped, and gain migratory and invasive capacity. EMT and the769reverse of this program, termed mesenchymal-to-epithelial tran-770sition (MET), induce multiple fundamental changes in cell phys-771iology in addition to the morphologic differences. EMT can be772triggered by a set of pleiotropically acting transcription factors773(TF), including Snail1/2, Twist1/2, and Zeb1/2 (37). Snail774(Snail1), as one of the most important EMT-TFs, is frequently775overexpressed in different types of metastatic cancers, including776colorectal carcinoma (38). Epigenetic modification, transcrip-777tional, and posttranscriptional regulatory mechanisms are778involved tightly in Snail expression (39). Our previous study779showed that SATB2 could reduce Snail protein level, induce MET,780and suppress colorectal carcinoma metastasis (16). However, the

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

www.aacrjournals.org Cancer Res; 2019 13

783 mechanism involved in the repression of Snail by SATB2 has not784 been clearly defined.785 As a MAR-bind protein, SATB2 has been reported to bind AT-786 rich regulatory elements for transcriptional regulation in pre-B787 cells, osteoblasts, brain, and embryonic stem cells (26, 40, 41).788 Here, we further identified that SATB2 recruited co-repressor789 HDAC1 to Snail promoter and removed activating histone acet-790 ylation marks, thereby decreasing Snail expression. Consistent791 with our results, HDAC inhibitor treatment enhances Snail acet-792 ylation and induces EMT via inducing Snail transcription (42). In793 the first half of present study, we found SATB2-AS1 could recruit794 p300. Although, p300 occupancy at Snail promoter was observed,795 SATB2-AS1-overexpression could not significantly increase this796 occupancy ability. Thus, the Snail expression did not appear to be797 directly regulated by SATB2-AS1-recruited p300. It is likely that798 there are other factors that recruitment p300 to regulate Snail799 expression. To our knowledge, we demonstrated a novel mech-800 anism responsible for the transcriptional regulation of Snail801 mediated by SATB2, a MAR-bind protein. Our study furthermore802 revealed that both upregulation of SATB2-AS1 and SATB2803 repressed Snail expression, thereby increasing epithelial markers804 and decreasing mesenchymal markers, and induced MET, which805 explained the mechanism of SATB2-AS1 inhibited colorectal806 carcinoma progression. Strikingly, Snail could decrease the807 expression of SATB2-AS1. These data hinted the probable exis-808 tence of a negative feedback loopbetween SATB2-AS1 and Snail in809 colorectal carcinoma cells. SATB2-AS1 transcriptional inactiva-810 tion and the associationbetween the SATB2-AS1 andSnail present811 an interesting issue for further investigation.812 The published study showed that Snail impairs the transition813 from early to late G1 by maintaining low levels of Cyclin D and814 blocks the G1–S transition by maintaining high levels of p21 in815 MDCK and MCA3D cells after 72 hours in culture with basal816 conditions (43). However, Snail-overexpressing cells can respond817 to mitogenic signals by transiently decreasing p21 expression,818 which favors the transition to S phase (43). The study also819 reported that Snail repressed p21 expression in infected B cells820 contributing to KSHV induced B-cell oncogenesis (44). Snail821 knockdown was reported to led to the upregulation of p21 and822 downregulation of Cdc2, causing a decrease of G1 and S periods823 and an increase of the G2–M population (45). In our rescue824 experiments, we observed that Snail overexpressing promoted825 G1–S transition and blocked the phenotype caused by the over-826 expression of SATB2-AS1. These data indicated that Snail has827 complicated roles on regulating cell-cycle progression depending828 on cell lines, cell culture conditions, and its target genes in829 different cancer cells. In addition, SATB2-AS1 might directly/

831indirectly regulate a number of genes contributing globally to832cancer progression, including cell-cycle regulation. Snail is only833one of many genes regulated by SATB2-AS1. Further characteri-834zation of these genes will provide new insights in SATB2-AS1-835mediated cell-cycle regulation.836In summary, we determined that SATB2-AS1 was downregu-837lated in colorectal carcinoma and consequently decreased p300838recruitment. In turn, the reduction in p300 recruitment sup-839pressed the accumulation of the active marks H3K27ac and840H3K9ac and repressed SATB2 levels, then activating Snail expres-841sion and inducing EMT to promote colorectal carcinoma cell842proliferation, invasion and migration in vivo and in vitro. Here,843we reported the involvement of SATB2-AS1 as an epigenetic844regulator that modulates CRC aggressiveness by regulating of845SATB2 expression and EMT. Overall, we highlight a novel mech-846anism of regulation by the SATB2-AS1-SATB2-Snail axis, which is847mediated by epigenetic regulation. Our results suggest that848SATB2-AS1 deregulation contributes to the development and849progression of colorectal carcinoma, suggesting that SATB2-AS1850may represent an effective target for blocking or reversing colo-851rectal carcinoma progression.

852Disclosure of Potential Conflicts of Interest853No potential conflicts of interest were disclosed Q8.

854Authors' Contributions855Conception and design: Y.-Q. Ding, S. Wang856Acquisition of data (provided animals, acquired and managed patients,857provided facilities, etc.): Y.-Q. Wang, D.-M. Jiang, S.-S. Hu, Y. Han858Analysis and interpretation of data (e.g., statistical analysis, biostatistics,859computational analysis): Y.-Q. Wang, D.-M. Jiang, S.-S. Hu, L. Zhao, L. Wang,860M.-L. Ai, H.-J. Jiang861Writing, review, and/or revision of the manuscript: Y.-Q. Wang, D.-M. Jiang,862S. Wang863Administrative, technical, or material support (i.e., reporting or organizing864data, constructing databases): L. Zhao, M.-H. Yang,865Study supervision: Y.-Q. Ding, S. Wang Q9

866Acknowledgments867The National Basic Research Program of China (973 Program,8682015CB554002), the National Natural Science Foundation of China869(81472318 and81502483), and theNatural Science Foundation ofGuangdong870Province (2014A030313308 and 2014A030310134) supported this work.

871The costs of publication of this articlewere defrayed inpart by the payment of872page charges. This article must therefore be hereby marked advertisement in873accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

874Received September 16, 2018; revised January 17, 2019; accepted March 4,8752019; published first xx xx, xxxx.

876 References877 1. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer878 statistics, 2012. CA Cancer J Clin 2015;65:87–108.879 2. Massague J, Obenauf AC. Metastatic colonization by circulating tumour880 cells. Nature 2016;529:298–306.881 3. GuttmanM,Donaghey J, Carey BW, GarberM, Grenier JK,MunsonG, et al.882 lincRNAs act in the circuitry controlling pluripotency and differentiation.883 Nature 2011;477:295–300.884 4. Mercer TR, Dinger ME, Mattick JS. Long non-coding RNAs: insights into885 functions. Nat Rev Genet 2009;10:155–9.886 5. Yuan JH, Yang F, Wang F, Ma JZ, Guo YJ, Tao QF, et al. A long noncoding887 RNA activated by TGF-beta promotes the invasion-metastasis cascade in888 hepatocellular carcinoma. Cancer Cell 2014;25:666–81.

8906. Huang JF, Guo YJ, Zhao CX, Yuan SX, Wang Y, Tang GN, et al. Hepatitis B891virus X protein (HBx)-related long noncoding RNA (lncRNA) down-892regulated expression by HBx (Dreh) inhibits hepatocellular carcinoma893metastasis by targeting the intermediate filament protein vimentin.894Hepatology 2013;57:1882–92.8957. Yang F, Zhang L, Huo XS, Yuan JH, Xu D, Yuan SX, et al. Long noncoding896RNA high expression in hepatocellular carcinoma facilitates tumor growth897through enhancer of zeste homolog 2 in humans. Hepatology 2011;54:8981679–89.8998. Gupta RA, Shah N, Wang KC, Kim J, Horlings HM, Wong DJ, et al. Long900non-coding RNA HOTAIR reprograms chromatin state to promote cancer901metastasis. Nature 2010;464:1071–6.

Wang et al.

Cancer Res; 2019 Cancer Research14

904 9. Zhang YF, Liu L, Ding YQ. Isolation and characterization of human905 colorectal cancer cell subline with unique metastatic potential in the liver.906 Nan Fang Yi Ke Da Xue Xue Bao 2007;27:126–30.907 10. Wang S, Zhou J, Wang XY, Hao JM, Chen JZ, Zhang XM, et al. Down-908 regulated expression of SATB2 is associated with metastasis and poor909 prognosis in colorectal cancer. J Pathol 2009;219:114–22.910 11. Moyer MP, Manzano LA, Merriman RL, Stauffer JS, Tanzer LR. NCM460, a911 normal human colon mucosal epithelial cell line. In Vitro Cell Dev912 Biol Anim 1996;32:315–7.913 12. Sun M, Nie F, Wang Y, Zhang Z, Hou J, He D, et al. LncRNA HOXA11-AS914 promotes proliferation and invasion of gastric cancer by scaffolding the915 chromatin modification factors PRC2, LSD1, and DNMT1. Cancer Res916 2016;76:6299–310.917 13. Zhang F, Li K, Pan M, Li W, Wu J, Li M, et al. miR-589 promotes gastric918 cancer aggressiveness by a LIFR-PI3K/AKT-c-Jun regulatory feedback loop.919 J Exp Clin Cancer Res 2018;37:152Q10 .920 14. Xing Z, Lin A, Li C, Liang K,Wang S, Liu Y, et al. lncRNA directs cooperative921 epigenetic regulation downstream of chemokine signals. Cell 2014;159:922 1110–25.923 15. Lin A, Li C, Xing Z,HuQ, Liang K,Han L, et al. The LINK-A lncRNA activates924 normoxic HIF1alpha signalling in triple-negative breast cancer. Nat Cell925 Biol 2016;18:213–24.926 16. Yang MH, Yu J, Jiang DM, Li WL, Wang S, Ding YQ. microRNA-182 targets927 special AT-rich sequence-binding protein 2 to promote colorectal cancer928 proliferation and metastasis. J Translat Med 2014;12:109Q11 .929 17. Yang MH, Yu J, Chen N, Wang XY, Liu XY, Wang S, et al. Elevated930 microRNA-31 expression regulates colorectal cancer progression by repres-931 sing its target gene SATB2. PLoS One 2013;8:e85353.932 18. Liu C, Lin J. Long noncoding RNA ZEB1-AS1 acts as an oncogene in933 osteosarcoma by epigenetically activating ZEB1. Am J Translat Res 2016;934 8:4095–105.935 19. Krawczyk M, Emerson BM. p50-associated COX-2 extragenic RNA936 (PACER) activates COX-2 gene expression by occluding repressive937 NF-kappaB complexes. Elife 2014;3:e01776.938 20. Gyorgy AB, SzemesM, de Juan RomeroC, Tarabykin V, AgostonDV. SATB2939 interacts with chromatin-remodeling molecules in differentiating cortical940 neurons. Eur J Neurosci 2008;27:865–73.941 21. Iyer MK, Niknafs YS, Malik R, Singhal U, Sahu A, Hosono Y, et al. The942 landscapeof longnoncodingRNAs in the human transcriptome.NatGenet943 2015;47:199–208.944 22. Yan X, Hu Z, Feng Y, Hu X, Yuan J, Zhao SD, et al. Comprehensive genomic945 characterization of long non-coding RNAs across human cancers.946 Cancer Cell 2015;28:529–40.947 23. Bhan A, Mandal SS. LncRNA HOTAIR: a master regulator of chromatin948 dynamics and cancer. Biochim Biophys Acta 2015;1856:151–64.949 24. Ma Y, Yang Y, Wang F, Moyer MP, Wei Q, Zhang P, et al. Long non-coding950 RNA CCAL regulates colorectal cancer progression by activating Wnt/beta-951 catenin signalling pathway via suppression of activator protein 2alpha.952 Gut 2016;65:1494–504.953 25. Liu SH, Zhu JW,XuHH,ZhangGQ,WangY, Liu YM, et al. A novel antisense954 long non-coding RNA SATB2-AS1 overexpresses in osteosarcoma and955 increases cell proliferation and growth. Mol Cell Biochem 2017;430:956 47–56.957 26. Dobreva G, ChahrourM,DautzenbergM, Chirivella L, Kanzler B, Farinas I,958 et al. SATB2 is amultifunctional determinant of craniofacial patterning and959 osteoblast differentiation. Cell 2006;125:971–86.960 27. Faghihi MA, Modarresi F, Khalil AM, Wood DE, Sahagan BG, Morgan TE,961 et al. Expression of a noncoding RNA is elevated in Alzheimer's disease and962 drives rapid feed-forward regulation of beta-secretase. Nat Med 2008;14:963 723–30.

96528. Beckedorff FC, Ayupe AC, Crocci-Souza R, Amaral MS, Nakaya HI,966Soltys DT, et al. The intronic long noncoding RNA ANRASSF1 recruits967PRC2 to the RASSF1A promoter, reducing the expression of RASSF1A968and increasing cell proliferation. PLos Genet 2013;9:e1003705.96929. Banfai B, Jia H, Khatun J, Wood E, Risk B, Gundling WE Jr., et al. Long970noncoding RNAs are rarely translated in two human cell lines. Genome Res9712012;22:1646–57.97230. ChenQ, Cai J, Wang Q,Wang Y, LiuM, Yang J, et al. Long Noncoding RNA973NEAT1, Regulated by the EGFR pathway, contributes to glioblastoma974progression through the WNT/beta-catenin pathway by scaffolding EZH2.975Clin Cancer Res 2018;24:684–95.97631. Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and977evolving paradigms. Cell 2011;147:275–92.97832. Shibue T, Weinberg RA. EMT, CSCs, and drug resistance: the mechanistic979link and clinical implications. Nat Rev Clin Oncol 2017;14:611–29.98033. Krebs AM,Mitschke J, Lasierra LosadaM, SchmalhoferO, BoerriesM, Busch981H, et al. The EMT-activator Zeb1 is a key factor for cell plasticity and982promotes metastasis in pancreatic cancer. Nat Cell Biol 2017;19:518–29.98334. Fan F, Samuel S, Evans KW, Lu J, Xia L, ZhouY, et al.Overexpressionof snail984induces epithelial-mesenchymal transition and a cancer stem cell-like985phenotype in human colorectal cancer cells. Cancer Med 2012;1:5–16.98635. Zhu W, Cai MY, Tong ZT, Dong SS, Mai SJ, Liao YJ, et al. Overexpression987of EIF5A2 promotes colorectal carcinoma cell aggressiveness by upregulat-988ing MTA1 through C-myc to induce epithelial-mesenchymaltransition.989Gut 2012;61:562–75.99036. Hwang WL, Yang MH, Tsai ML, Lan HY, Su SH, Chang SC, et al. SNAIL991regulates interleukin-8 expression, stem cell-like activity, and tumorige-992nicity of human colorectal carcinoma cells. Gastroenterology 2011;141:993279–91.99437. Li GY, Wang W, Sun JY, Xin B, Zhang X, Wang T, et al. Long non-coding995RNAs AC026904.1 and UCA1: a "one-two punch" for TGF-beta-induced996SNAI2 activation and epithelial-mesenchymal transition in breast cancer.997Theranostics 2018;8:2846–61.99838. Feng M, Feng J, Chen W, Wang W, Wu X, Zhang J, et al. Lipocalin2999suppresses metastasis of colorectal cancer by attenuating NF-kappaB-1000dependent activation of snail and epithelial mesenchymal transition.1001Mol Cancer 2016;15:77 Q12.100239. XuW, Liu H, Liu ZG, Wang HS, Zhang F, Wang H, et al. Histone deacetylase1003inhibitors upregulate Snail via Smad2/3phosphorylation and stabilizationof1004Snail to promote metastasis of hepatoma cells. Cancer Lett 2018;420:1–13.100540. Dobreva G, Dambacher J, Grosschedl R. SUMO modification of a novel1006MAR-binding protein, SATB2, modulates immunoglobulin mu gene1007expression. Genes Develop 2003;17:3048–61.100841. Savarese F, Davila A, Nechanitzky R, De La Rosa-Velazquez I, Pereira CF,1009Engelke R, et al. Satb1 and Satb2 regulate embryonic stem cell differen-1010tiation and Nanog expression. Genes Develop 2009;23:2625–38.101142. Jiang GM, Wang HS, Zhang F, Zhang KS, Liu ZC, Fang R, et al. Histone1012deacetylase inhibitor induction of epithelial-mesenchymal transitions via1013up-regulation of Snail facilitates cancer progression. Biochim Biophys Acta10142013;1833:663–71.101543. Vega S,Morales AV,OcanaOH, Valdes F, Fabregat I, NietoMA. Snail blocks1016the cell cycle and confers resistance to cell death. Genes Develop 2004;18:10171131–43.101844. Jha HC, Sun Z, Upadhyay SK, El-Naccache DW, Singh RK, Sahu SK, et al.1019KSHV-mediated regulation of Par3 and SNAIL contributes to B-cell pro-1020liferation. PLoS Pathog 2016;12:e1005801.102145. Yang X, Han M, Han H, Wang B, Li S, Zhang Z, et al. Silencing Snail1022suppresses tumor cell proliferation and invasion by reversing epithelial-to-1023mesenchymal transition and arresting G2–M phase in non-small cell lung1024cancer. Int J Oncol 2017;50:1251–60.

www.aacrjournals.org Cancer Res; 2019 15

SATB2-AS1/SATB2/Snail Axis Inhibits CRC Aggressiveness

AUTHOR QUERIES

AUTHOR PLEASE ANSWER ALL QUERIES

Q1: Page: 1: Author: Per journal style, genes, alleles, loci, and oncogenes are italicized;proteins are roman. Please check throughout to see that the words are styled correctly.AACR journals have developed explicit instructions about reporting results from experi-ments involving the use of animalmodels as well as the use of approved gene and proteinnomenclature at their first mention in the manuscript. Please review the instructions athttp://aacrjournals.org/content/authors/editorial-policies#genenomen to ensure thatyour article is in compliance. If your article is not in compliance, please make theappropriate changes in your proof.

Q2: Page: 1: Author: Please verify the drug names and their dosages used in the article.

Q3: Page: 1: Author: Please verify the affiliations and their corresponding author links.

Q4: Page: 1: Author: Please verify the corresponding author details.

Q5: Page: 1: Author: Note that AACR journals do not use such abbreviations as "BC" for"breast cancer," "CRC" for colorectal cancer, or "PCa" for "prostate cancer," or other ad hoctwo-word abbreviations for diseases that incorporate a part of the body. However, theseabbreviations are acceptable in tables or figures where space is an issue as long as they aredefined in the legend or table notes.

Q6: Page: 3: Author: Units ofmeasurement have been changed here and elsewhere in the textfrom "M" to "mol/L," and related units, such as "mmol/L" and "mmol/L," in figures,legends, and tables in accordance with journal style, derived from the Council of ScienceEditors Manual for Authors, Editors, and Publishers and the Syst�eme internationald'unit�es. Please note if these changes are not acceptable or appropriate in this instance.

Q7: Page: 4: Author: Please confirmquality/labeling of all images includedwithin this article.Thank you.

Q8: Page: 14: AU:/PE: The conflict-of-interest disclosure statement that appears in the proofincorporates the information from forms completed and signed off on by each individualauthor. No factual changes can be made to disclosure information at the proof stage.However, typographical errors or misspelling of author names should be noted on theproof and will be corrected before publication. Please note if any such errors need to becorrected. Is the disclosure statement correct?

Q9: Page: 14: Author: The contribution(s) of each author are listed in the proof under theheading "Authors' Contributions." These contributions are derived from forms complet-ed and signed off on by each individual author. As the corresponding author, you arepermitted to make changes to your own contributions. However, because all authorssubmit their contributions individually, you are not permitted to make changes in thecontributions listed for any other authors. If you feel strongly that an error is beingmade,then you may ask the author or authors in question to contact us about making the

changes. Please note, however, that themanuscript would be held from further processinguntil this issue is resolved.

Q10: Page: 15: Author: Please provide last page range for ref. 13.

Q11: Page: 15: Author: Please provide last page range for ref. 16.

Q12: Page: 15: Author: Please provide last page range for ref. 38.

AU: Below is a summary of the name segmentation for the authors according to our records.The First Name and the Surname data will be provided to PubMed when the article is indexedfor searching. Please check each name carefully and verify that theFirstName andSurname arecorrect. If a name is not segmented correctly, please write the correct First Name and Surnameon this page and return it with your proofs. If no changes are made to this list, we will assumethat the names are segmented correctly, and the names will be indexed as is by PubMed andother indexing services.

First Name Surname

Yi-Qing Wang

Dong-Mei Jiang

Sha-Sha Hu

Li Zhao

Lan Wang

Min-hui Yang

Mei-Ling Ai

Hui-Juan Jiang

Yue Han

Yan-Qing Ding

Shuang Wang