Introduction
Epigenetics is the study of reversible gene expression and heritable variation that not caused by changes in DNA sequence (Jones and Baylin, 2002; Bird, 2002; Lee, 2012; Suva et al., 2013). The best example of epigenetic changes is cell differentiation in eukaryotes (Bird, 2002; Reik et al., 2001; Li, 2002; Reik, 2007). In the process of morphogenesis, a variety of pluripotent stem cells in embryo are derived from totipotent stem cells, which can further differentiate into different cells. The biological process can be finished by activing some genes and suppressing others. The phenomenon of epigenetics include DNA methylation, genomic imprinting, etc (Egger et al., 2004; Jones and Takai, 2001).

DNA methylation is one of the hottest studied epigenetics modification in mammals, which control gene expression and silencing in normal cells (Li et al., 1993b; Li et al., 1993a; Bird, 2002; Jones and Taylor, 1980; Jones, 2012; Greer and Shi, 2012). It is associated with histone modification. For the interaction among kinds of epigenetics modification is extremely important to the process which regulating the function of chromatin by the change of chromatic structure (Bird, 1980). Methyl group can bind with CpG dinucleotide cytosine by covalent bonds, we named the loci gathered a large number of CPG dinucleotide as CPG island. Methyltransferase play a role in formation and maintenance of methylation patterns (Bird, 1986; Santos et al., 2002).

The changes of DNA methylation pattern has been observed in cancer cells (Robertson, 2001; Esteller and Herman, 2002; Rountree et al., 2001; Baylin et al., 2001; Momparler and Bovenzi, 2000; Das and Singal, 2004). Current researches suggested that high or low methylation is found in different locations, and DNA methylation play a role in the mutation of cancer (Ehrlich, 2002; Esteller et al., 2001; Jones, 1996).

IGFBP2 (insulin-like growth factor-binding protein 2) is one of the family of ISGBP which binding different kinds of IGFs. It inhibits the regulation function of IGFs in growth and development (Fisher et al., 2005; Heald et al., 2006). IGFBP prolong the half-life of IGFs and inhibit or stimulate the growth function of IGFs. They interact with IGFs through the receptor on the surface of cell.

IGFBP-2 mRNA is already expressed in preimplantation embryo (Prelle et al., 2001), and expression continues at high levels in many tissues during embryonic and fetal development (Schuller et al., 1993; van Kleffens et al., 1998). In the postnatal period, IGFBP-2 is the second most abundant IGFBP in the circulation and is present in various other biological fluids and tissues of many vertebrate species (Blum et al., 1993; Hwa et al., 1999).

Clinical researches discover that patients with cancer of high expression of IGFBP2 often have a shorter survival time (Busund et al., 2005; Lin et al., 2009; Fukushima and Kataoka, 2007; Hsieh et al., 2010). This study sought for the epigenetic regulation elements of IGFBP2, and then used it to the survival analysis. We obtained the epigenetic regulation elements that could separate patients with the long survival time and patients with short survival time. These epigenetic regulation elements can be used for the clinical study of drug development for the treatment of cancer.

1 Method
1.1 Datasets
We obtained colon cancer data from TCGA (https://tcga-data.nci.nih.gov/tcga/) including 450 k DNA methylation data, gene expression data and clinical information data. The gene expression data includes 273 cancer samples and 41 normal samples from UNC IlluminaHiSeq RNASeqV2 level3 data. The 450k DNA methylation data includes 339 cancer samples and 38 normal samples from JUH-USC Human Methylation 450 level3 data. We obtained samples with both DNA methylation data and gene expression data, including 266 cancer samples and 19 normal samples.

1.2 Differentially methylation analysis
IGFBP2 gene expression data was abstracted, and divided all the cancer samples into two groups: high expression group and low expression group. SAM （Significance Analysis of Microarrays） is a statistics tools for searching remarkable genes in microarray data sets. It was used to distinguish differentially methylated site in both high and low gene expression group of IGFBP2.
 
1.3 The screening of methylation sites and survival analysis
The relationship between IGFBP2 gene expression and differentially methylated sites was obtained from the Pearson correlation coefficient. Through 1000 permutation Pearson correlation coefficient, the DNA methylation sites that highly correlated with IGFBP2 gene expression was obtained. Then the chromosome coordinate of highly correlated DNA methylation sites was obtained to get the DNA methylation sites in cis regulation region of IGFBP2.

The COX regression analysis was used to get the DNA methylation sites that correlated with survival time. Then we used these DNA methylation sites to plot survival curve.

2 Results
2.1 Data processing and analysis
To filter the epigenetic regulation elements, we obtained colon adenocarcinoma data from TCGA (The Cancer Genome Atlas) with 450k DNA methylation data, RNA-seq data and the clinical information. All the data were level 3 that had been preprocessed by the TCGA. There were 314 samples in RNA-seq data include 273 cancer samples and 41 normal samples, whereas 301 cancer samples and 38 normal samples included in DNA methylation data. We filtered 285 samples with corresponding RNA-seq and DNA methylation data.

We drew the IGFBP2 expression value from RNA-seq and grouped the cancer samples into high expression group and low expression group through the mean value of IGFBP2 expression. There were 100 cancer samples in the high expression group and 166 cancer samples in the low expression group, besides 19 normal samples.

2.2 Differentially methylation sites
We filtered differentially methylated sites between the high expression group or low expression group and the normal sample group by SAM (Table 1). In the high expression group, we filtered 78942 high differentially methylated sites and 103567 low differentially methylated sites. In the low expression group, we filtered 18504 high differentially methylated sites and 73443 low differentially methylated sites.
 

2.3 Permutation
To get the highly correlative DNA methylation sites, Pearson's correlation had been calculated between the differentially methylated sites and the IGFBP2 expression. Permutation also was performed to get the 0.95 confidence interval of significantly correlative DNA methylation sites. After 1000 times of random perturbation, the distribution of correlation coefficients followed Gaussian distributions (Figure 1).
 

We obtained 39970 and 62417 highly correlative differentially DNA methylation sites in highly differentially methylated group and lowly differentially methylated group. Correspondingly, in the low expression group, we obtained 6997 and 20300 highly correlative differentially DNA methylation. The correlation coefficient of each group had been show in Figure 2-3. We could acquire that highly correlative differentially DNA methylation sites in high expression group were more than in low expression group.
 

 
2.4 Obtaining CpG sites in cis regulation region
In order to obtain the DNA methylation sites in the cis regulation region, we get the chromosome coordinate of each highly correlative differentially DNA methylation sites. Because the IGFBP2 was on the second chromosome between 216449551 and 217498127, we obtained the DNA methylation sites in the region of 1Mb to the 216449551 (Table 2).
 

 
2.5 Survival analysis
These DNA methylation sites with 25 from high expression group and 9 from low expression group had been found in the cis regulation region, but we were not sure if there really regulated the expression of IGFBP2. Therefore, we analysed these DNA methylation sites with the corresponding survival time through cox regression analysis. The cg09410607 and cg22954687 had been show significant correlation with survival time. Then we plotted Kaplan-Meier survival curve with the two DNA methylation site and the survival time (Figure 4). We concluded that the cg09410607 and cg22954687 could separate the sample of long survival time and short survival time, in addition, they could also regulate the expression of IGFBP2. The p-value of the survival curve is 0.035.
 

3 Discussion
The overexpression of IGFBP2 has been observed in many kinds of cancer, such as breast cancer, colorectal cancer, neuroendocrine cancer, etc (Busund et al., 2005; Mishra et al., 1998; Yazawa et al., 2009). It is one of the most highly expressed IGFBPs in neuroblastomas, glial tumors, and prostate cancers (Menouny et al., 1997; Sallinen et al., 2000; Fuller et al., 1999; Cindolo et al., 2007; Yazawa et al., 2009). Mechanisms of its overexpression have been investigated in many respects, but few are from the epigenetic field (Yazawa et al., 2009).

This study investigated the epigenetic regulation of the expression of IGFBP2. Through the differentially DNA methylation analysis and the correlation analysis, we obtained highly correlative DNA methylation sites in the cis regulation region of IGFBP2, and they significantly correlated with survival time.

The result showed that the high expression of IGFBP2 in cancer patients often indicated short survival time. In the high expression group of IGFBP2, the difference of DNA methylation state increased. On the other hand, the expression of IGFBP2 was regulated by epigenetic elements including cg09410607 and cg22954687. The two DNA methylation sites could significantly separate the sample of long survival time and short survival time.

Acknowledgments  
This work was supported by the Science Innovation Project (grants 2015003) and the Innovation and Technology
special Fund for excellent academic leader of Harbin (grant number 2015RAXYJ051).

References
Baylin S.B., Esteller M., Rountree M.R., Bachman K.E., Schuebel K., and Herman J.G., 2001, Aberrant patterns of DNA methylation, chromatin formation and gene expression in cancer, Hum Mol Genet, 10: 687-692
http://dx.doi.org/10.1093/hmg/10.7.687

Bird A., 2002, DNA methylation patterns and epigenetic memory, Genes Dev, 16: 6-21
http://dx.doi.org/10.1101/gad.947102

Bird A.P., 1980, DNA methylation and the frequency of CpG in animal DNA, Nucleic Acids Res, 8: 1499-1504
http://dx.doi.org/10.1093/nar/8.7.1499
 
Bird A.P., 1986, CpG-rich islands and the function of DNA methylation, Nature, 321: 209-213
http://dx.doi.org/10.1038/321209a0

Blum W.F., Horn N., Kratzsch J., Jorgensen J.O., Juul A., Teale D., Mohnike K., and Ranke M.B., 1993, Clinical studies of IGFBP-2 by radioimmunoassay, Growth Regul, 3: 100-104

Busund L.T., Richardsen E., Busund R., Ukkonen T., Bjornsen T., Busch C., and Stalsberg H., 2005, Significant expression of IGFBP2 in breast cancer compared with benign lesions, J Clin Pathol, 58: 361-366
http://dx.doi.org/10.1136/jcp.2004.020834

Cindolo L., Franco R., Cantile M., Schiavo G., Liguori G., Chiodini P., Salzano L., Autorino R., Di Blasi A., Falsaperla M., Feudale E., Botti G., Gallo A., and Cillo C., 2007, NeuroD1 expression in human prostate cancer: can it contribute to neuroendocrine differentiation comprehension?, Eur Urol, 52: 1365-1373
http://dx.doi.org/10.1016/j.eururo.2006.11.030

Das P.M., and Singal R., 2004, DNA methylation and cancer, J Clin Oncol, 22: 4632-4642
http://dx.doi.org/10.1200/JCO.2004.07.151

Egger G., Liang G., Aparicio A., and Jones P.A., 2004, Epigenetics in human disease and prospects for epigenetic therapy, Nature, 429: 457-463
http://dx.doi.org/10.1038/nature02625

Ehrlich M., 2002, DNA methylation in cancer: too much, but also too little, Oncogene, 21: 5400-5413
http://dx.doi.org/10.1038/sj.onc.1205651

Esteller M., Corn P.G., Baylin S.B., and Herman J.G., 2001, A gene hypermethylation profile of human cancer, Cancer Res, 61: 3225-3229

Esteller M., and Herman J.G., 2002, Cancer as an epigenetic disease: DNA methylation and chromatin alterations in human tumours, J Pathol, 196: 1-7
http://dx.doi.org/10.1002/path.1024

Fisher M.C., Meyer C., Garber G., and Dealy C.N., 2005, Role of IGFBP2, IGF-I and IGF-II in regulating long bone growth, Bone, 37: 741-750
http://dx.doi.org/10.1016/j.bone.2005.07.024

Fukushima T., and Kataoka H., 2007, Roles of insulin-like growth factor binding protein-2 (IGFBP-2) in glioblastoma, Anticancer Res, 27: 3685-3692

Fuller G.N., Rhee C.H., Hess K.R., Caskey L.S., Wang R., Bruner J.M., Yung W.K., and Zhang W., 1999, Reactivation of insulin-like growth factor binding protein 2 expression in glioblastoma multiforme: a revelation by parallel gene expression profiling, Cancer Res, 59: 4228-4232

Greer E.L., and Shi Y., 2012, Histone methylation: a dynamic mark in health, disease and inheritance, Nat Rev Genet, 13: 343-357
http://dx.doi.org/10.1038/nrg3173

Heald A.H., Kaushal K., Siddals K.W., Rudenski A.S., Anderson S.G., and Gibson J.M., 2006, Insulin-like growth factor binding protein-2 (IGFBP-2) is a marker for the metabolic syndrome, Exp Clin Endocrinol Diabetes, 114: 371-376
http://dx.doi.org/10.1055/s-2006-924320

Hsieh D., Hsieh A., Stea B., and Ellsworth R., 2010, IGFBP2 promotes glioma tumor stem cell expansion and survival, Biochem Biophys Res Commun, 397: 367-372
http://dx.doi.org/10.1016/j.bbrc.2010.05.145

Hwa V., Oh Y., and Rosenfeld R.G., 1999, The insulin-like growth factor-binding protein (IGFBP) superfamily, Endocr Rev, 20: 761-787
http://dx.doi.org/10.1210/er.20.6.761

Jones P.A., 1996, DNA methylation errors and cancer, Cancer Res, 56: 2463-2467

Jones P.A., 2012, Functions of DNA methylation: islands, start sites, gene bodies and beyond, Nat Rev Genet, 13: 484-492
http://dx.doi.org/10.1038/nrg3230

Jones P.A., and Baylin S.B., 2002, The fundamental role of epigenetic events in cancer, Nat Rev Genet, 3: 415-428

Jones P.A., and Takai D., 2001, The role of DNA methylation in mammalian epigenetics, Science, 293: 1068-1070
http://dx.doi.org/10.1126/science.1063852

Jones P.A., and Taylor S.M., 1980, Cellular differentiation, cytidine analogs and DNA methylation, Cell, 20: 85-93
http://dx.doi.org/10.1016/0092-8674(80)90237-8

Lee J.T., 2012, Epigenetic regulation by long noncoding RNAs, Science, 338: 1435-1439
http://dx.doi.org/10.1126/science.1231776

Li E., 2002, Chromatin modification and epigenetic reprogramming in mammalian development, Nat Rev Genet, 3: 662-673
http://dx.doi.org/10.1038/nrg887

Li E., Beard C., Forster A.C., Bestor T.H., and Jaenisch R., 1993a, DNA methylation, genomic imprinting, and mammalian development, Cold Spring Harb Symp Quant Biol, 58: 297-305
http://dx.doi.org/10.1101/SQB.1993.058.01.035

Li E., Beard C., and Jaenisch R., 1993b, Role for DNA methylation in genomic imprinting, Nature, 366: 362-365
http://dx.doi.org/10.1038/366362a0

Lin Y., Jiang T., Zhou K., Xu L., Chen B., Li G., Qiu X., Jiang T., Zhang W., and Song S.W., 2009, Plasma IGFBP-2 levels predict clinical outcomes of patients with high-grade gliomas, Neuro Oncol, 11: 468-476
http://dx.doi.org/10.1215/15228517-2008-114
 
Menouny M., Binoux M., and Babajko S., 1997, Role of insulin-like growth factor binding protein-2 and its limited proteolysis in neuroblastoma cell proliferation: modulation by transforming growth factor-beta and retinoic acid, Endocrinology, 138: 683-690

Mishra L., Bass B., Ooi B.S., Sidawy A., and Korman L., 1998, Role of insulin-like growth factor-I (IGF-I) receptor, IGF-I, and IGF binding protein-2 in human colorectal cancers, Growth Horm IGF Res, 8: 473-479
http://dx.doi.org/10.1016/S1096-6374(98)80300-6

Momparler R.L., and Bovenzi V., 2000, DNA methylation and cancer, J Cell Physiol, 183: 145-154
http://dx.doi.org/10.1002/(SICI)1097-4652(200005)183:2%3C145::AID-JCP1%3E3.0.CO;2-V

Prelle K., Stojkovic M., Boxhammer K., Motlik J., Ewald D., Arnold G.J., and Wolf E., 2001, Insulin-like growth factor I (IGF-I) and long R(3)IGF-I differently affect development and messenger ribonucleic acid abundance for IGF-binding proteins and type I IGF receptors in in vitro produced bovine embryos, Endocrinology, 142: 1309-1316
http://dx.doi.org/10.1210/en.142.3.1309

Reik W., 2007, Stability and flexibility of epigenetic gene regulation in mammalian development, Nature, 447: 425-432
http://dx.doi.org/10.1038/nature05918

Reik W., Dean W., and Walter J., 2001, Epigenetic reprogramming in mammalian development, Science, 293: 1089-1093
http://dx.doi.org/10.1126/science.1063443

Robertson K.D., 2001, DNA methylation, methyltransferases, and cancer, Oncogene, 20: 3139-3155
http://dx.doi.org/10.1038/sj.onc.1204341

Rountree M.R., Bachman K.E., Herman J.G., and Baylin S.B., 2001, DNA methylation, chromatin inheritance, and cancer, Oncogene, 20: 3156-3165
http://dx.doi.org/10.1038/sj.onc.1204339

Sallinen S.L., Sallinen P.K., Haapasalo H.K., Helin H.J., Helen P.T., Schraml P., Kallioniemi O.P., and Kononen J., 2000, Identification of differentially expressed genes in human gliomas by DNA microarray and tissue chip techniques, Cancer Res, 60: 6617-6622

Santos F., Hendrich B., Reik W., and Dean W., 2002, Dynamic reprogramming of DNA methylation in the early mouse embryo, Dev Biol, 241: 172-182
http://dx.doi.org/10.1006/dbio.2001.0501

Schuller A.G., Zwarthoff E.C., and Drop S.L., 1993, Gene expression of the six insulin-like growth factor binding proteins in the mouse conceptus during mid- and late gestation, Endocrinology, 132: 2544-2550

Suva M.L., Riggi N., and Bernstein B.E., 2013, Epigenetic reprogramming in cancer, Science, 339: 1567-1570
http://dx.doi.org/10.1126/science.1230184

Van Kleffens M., Groffen C., Lindenbergh-Kortleve D.J., Van Neck J.W., Gonzalez-Parra S., Dits N., Zwarthoff E.C., and Drop S.L., 1998, The IGF system during fetal-placental development of the mouse, Mol Cell Endocrinol, 140: 129-135
http://dx.doi.org/10.1016/S0303-7207(98)00041-0

Yazawa T., Sato H., Shimoyamada H., Okudela K., Woo T., Tajiri M., Ogura T., Ogawa N., Suzuki T., Mitsui H., Ishii J., Miyata C., Sakaeda M., Goto K., Kashiwagi K., Masuda M., Takahashi T., and Kitamura H., 2009, Neuroendocrine cancer-specific up-regulating mechanism of insulin-like growth factor binding protein-2 in small cell lung cancer, Am J Pathol, 175: 976-987
http://dx.doi.org/10.2353/ajpath.2009.081004