Novel SFRP2 DNA Methylation Profile Following Neoadjuvant Therapy in Colorectal Cancer Patients with Different Grades of BMI

The relationship between body weight and different cancers is now well-recognized and among such cancers, colorectal cancer (CRC) is reported most frequently. Our group recently published findings, through an epigenome-wide association study, suggesting that body mass index (BMI) could act as a relevant risk factor in the CRC. In addition, aberrant SFRP2 methylation is one of the major mechanisms for Wnt signaling activation in CRC. Conversely, neoadjuvant chemo-radiotherapy appears to alter the rectal cancer epigenome. This study was aimed to evaluate the effect of obesity, measured by BMI, on the methylation of SFRP2 in tumor samples of patients with CRC. Non-treated CRC patients and CRC patients treated with pre-operative neoadjuvant therapy from 2011 to 2013 were included and classified by BMI < 25.0 kg/m2 and BMI > 25.0 kg/m2. SFRP2 DNA methylation in tumor samples was measured by pyrosequencing. Our findings suggest a possible interaction between SFRP2 methylation levels and BMI in CRC tumor samples. The correlation of SFRP2 hypomethylation with an elevated BMI was stronger within the non-treated CRC patient group than within the treated CRC patient group. We have successfully demonstrated that the beneficial association of tumor SFRP2 hypomethylation is dependent on patient BMI in non-treated CRC, suggesting a possible tumor suppressor role for SFRP2 in overweight and obese patients. Additional studies of clinical pathologies would be necessary to strengthen these preliminary results.


Introduction
One of the biggest challenges facing biomedical science is finding promising diagnostic, prognostic, and predictive biomarkers of non-communicable diseases of multi-factorial origin, such as colorectal September 2013. One cohort group consisted of 53 overweight or obese patients (BMI > 25 kg/m 2 ), while 22 were non-obese participants (BMI < 25 kg/m 2 ). All participants underwent surgery with curative intention, by hemicolectomy, lower anterior resection with ileostomy (caused by a carcinoma of the CRC), followed by a total mesocolorectal excision. A formalin-fixed paraffin-embedded (FFPE) (10 sections of 14 µm) tumor area and an adjacent tumor-free area were obtained from all CRC patients.
Participants included in the study had primary CRC whose medical records/pathological examinations were complete. The exclusion criteria were patients with inflammatory bowel disease (Crohn´s disease or ulcerative colitis) and patients who had evidence of hereditary non-polyposis colorectal cancer or familial adenomatous polyposis. All participants were anonymized and gave their written informed consent. The study was performed in accordance with the Declaration of Helsinki and was approved by the Ethics Committees of Virgen de la Victoria University Hospital (registration number 0311/PI7) (Málaga, Spain).
The clinical-pathological parameters of the cohort groups were confirmed by reviewing the patient medical records and pathology files. The pathological diagnosis, tumor histology, and staging were performed according to the Classification of the "World Health Organization Classification of Tumours of the Digestive System" (2016) [24]. Rectal cancer patients were treated according to local protocols, after a routine workout, including a turaco abdominal scan and a pelvis MRI. Those classified as T3-4 or N 1-3 were proposed for neoadjuvant chemoradiation treatment with pelvic radiotherapy 50Gy (2 Gy/fraction) and concomitant administration of fluoropyrimidine-based chemotherapy, followed by total mesorectal excision in 6-8 weeks. Patients were sub-classified as treated (those that were designated for neoadjuvant therapy), while those who did not receive any pre-operative treatment were classified as non-treated.
Consequently, RNA from cell lines was isolated using QiampRNA Mini Kit (Qiagen GmbH, Hilden, Germany), according to the manufacturer's instructions. The purity and concentration of the RNA was determined by the 260/280 absorbance ratio using the Nanodrop 2000 platform (Thermo scientific, USA). For cDNA synthesis, a fixed amount of 1 µg of the total RNA was reverse transcribed using random hexamers as primers and Moloney Murine Leukemia Virus reverse transcriptase (Roche Diagnostic, Rotkreuz, Switzerland). Gene expression was assessed by real-time PCR, using an Applied Biosystems 7500 Fast Real-Time PCR System (Applied Biosystems, Darmstadt, Germany) with TaqMan technology. The reaction was performed following the manufacturer's protocol of TaqMan technologies Premix Ex Taq™ (Probe qPCR) (Takara, Madrid, Spain), in a final volume of 12.5 µL. The cycle program consisted of an initial denaturing of 20 s at 95 • C, then 45 cycles of a 3 s denaturing phase at 95 • C, followed by a 30 s annealing period. The commercially available and revalidated TaqMan primer/probe sets used in our samples were SFRP2 and PPIA TaqMan Probes (IDT technologies, Madrid, Spain). A threshold cycle (Ct value) was obtained for each amplification curve and a ∆Ct value was first calculated by subtracting the Ct value for Cyclophilin A cDNA from the Ct value for each sample and transcript. Fold changes compared with the endogenous control were then determined by calculating 2 −∆Ct , and relative quantification results were obtained.

DNA Extraction, Bisulfite Treatment, and Pyrosequencing
Total genomic DNA from 75 CRC paraffin samples (formalin-fixed paraffin embedded (FFPE) (10 sections of 14 µm) was isolated using the QIAmp DNA FFPE Tissue Kit (Qiagen GmbH, Hilden, Germany), with a xylene wash, to remove the paraffin. DNA concentration and quality was determined using NanoDrop 2000 (Thermo scientific, USA). DNA methylation analyses were performed using bisulfite-treated DNA, followed by a highly quantitative analysis, based on PCR-based pyrosequencing. The bisulfite conversion was conducted with 2 µg of genomic DNA isolated from each sample, using Epitect Bisulfite conversion, according to the manufacturer´s instructions (Qiagen GmbH, Hilden, Germany). Consequently, the SFRP2 promoter and LINE-1 amplification steps were performed in a total volume of 25 µL, with a starting primer concentration of 10 µM (SFRP2: Primer 5 3 forward and reverse respectively: TTTGATTTTTTTAYGGTATTGGGGAGTA and ATAAAACCCRAAACCTACCC. LINE-1: Primer 5 3 forward and reverse respectively: TAGGGAGTGTTAGATAGTGG and AACTCCCTAACCCCTTAC). The reverse primer was biotinylated, in order to purify the final PCR product, using sepharose beads. Finally, 20 µL of the PCR products were pyro sequenced using the PyroMarkTMQ96 ID Pyrosequencing System, using a 0.4 µM sequencing primer. The primer sequences used in this analysis were designed using Qiagen's PyroMark Assay Design 2.0 software (Qiagen, Hilden, Germany).
The methylation level was expressed as the percentage of methylated cytosine over the sum of methylated and unmethylated cytosines. Non-CpG cytosine residues were used as built-in controls to verify bisulfite conversion. The values are expressed as the mean for all the sites and individually for seven CpGs at the SFRP2 gene promoter and for six CpGs at the LINE-1 sequence [27,28]. We also included unmethylated and methylated DNA as controls in each run (New England Biolabs, Ipswich, MA, USA).

Statistical Analysis
The results were expressed as the mean ± standard deviation (or percentage) where appropriate. Student's t test was used for comparing anthropometric and biochemical data between non-obese and overweight/obese groups. Paired t-test was performed to compare means between tumor and tumor-free area. Student's t test and Mann-Whitney U-test were used for comparisons of LINE-1 and SFRP2 methylation between study groups. Spearman's correlation analyses were performed to study the correlations between LINE-1 and SFRP2 methylation. A regression linear model was used to perform multivariate analysis. All analyses were performed using the R statistical software, version 2.8.1 (Department of Statistics, University of Auckland, Auckland, NZ; http://www.rproject.org/).

Clinical, Anthropometrical, and Biochemical Characteristics of the CRC Patients
There were no differences in anthropometric and biochemical variables between subjects with BMI < 25 or BMI > 25, except for the BMI itself (as expected) and the insulin levels (Table 1).

Methylation Status of SFRP2 in CRC Tissue
First of all, we initiated our analysis based on previous data of the SFRP2 promoter performed and published by our group and obtained from the Infinium Human Methylation 450 BeadChip array [14].The study sequence of the SFRP2 promoter used in our analysis showed seven CpG sites in the position −1500, relative to the expected transcription start site (+1),which was consequently utilized to determine the methylation status of the SFRP2 promoter ( Figure 1a).
We aimed to evaluate the SFRP2 methylation status in both CRC tissue and adjacent tumor-free tissue from 75 CRC patients, by pyrosequencing after bisulfite treatment.
The results revealed hyper-methylation of the SFRP2 promoter, since SFRP2 methylation in tumor tissue was significantly higher (40.30%)in comparison to the levels obtained for adjacent tumor-free tissue areas (17.86%). Conversely, LINE-1 methylation was significantly decreased in the tumor area (57.38%), when compared to the tumor-free area (63.50%) (Figure 1b).
In addition, to test whether the hyper-methylation of SFRP2 promoter is functionally associated with a down-regulated expression of SFRP2 at the transcriptomic level, we performed in vitro demethylation assays, using different colorectal carcinoma cell lines.
Accordingly, we employed RT-qPCR assays to evaluate whether SFRP2 mRNA is differentially expressed, before and after treatment, with a methylation inhibitor. First, we treated cell lines for 72 h with 5 µM of AZA, a methylation inhibitor. Consequently, we quantified the mRNA expression of SFRP2 through RT-qPCR. The HTC116 control showed undetectable mRNA expression of SFRP2, while the HCT116 treated with AZA resulted in a significant and clear increase of mRNA SFRP2 expression, as shown in Figure 1c. However, despite measuring the SFRP2 gene expression in three different human colorectal carcinoma cell lines (HCT116, LoVo, and Caco-2), we only observed expression recovery of SFRP2 in HCT116, after the AZA treatment. The lack of results from the other cell lines could be explained by their inherent biological heterogeneity from these cell lines.
from the other cell lines could be explained by their inherent biological heterogeneity from these cell lines.

Baseline Characteristics of Colorectal Cancer Patients with Regards to their BMI Category
Clinical, pathological, and tumor molecular features according to the BMI in proximal colon cancer (cecum, ascending colon, and transverse colon), distal colon cancer (descending colon, and sigmoid colon), and rectal colon cancer are summarized in Table 2.
We observed that there were no differences between the two groups for any of the oncological variables studied.

Baseline Characteristics of Colorectal Cancer Patients with Regards to their BMI Category
Clinical, pathological, and tumor molecular features according to the BMI in proximal colon cancer (cecum, ascending colon, and transverse colon), distal colon cancer (descending colon, and sigmoid colon), and rectal colon cancer are summarized in Table 2. We observed that there were no differences between the two groups for any of the oncological variables studied.

Methylation of SFRP2 in CRC Tissue is Associated with BMI
In order to investigate whether adiposity affects DNA methylation levels, the methylation pattern of SFRP2 and LINE-1 was compared in tumor tissue and tumor-free area, in both BMI groups. Our results showed that the SFRP2 methylation was significantly lower in overweight/obese individuals (34.06%), compared to the non-obese individuals (51.24%) (Figure 2a), while LINE-1 methylation did not show any significant difference in BMI (56.68% for non-obese subjects and for overweight/obese individuals 57.69%) (Figure 2b). We also observed a significant negative correlation between global LINE-1 and SFRP2 methylation in CRC tumor tissue (r = −0.329, p = 0.011) (Figure 2c). However, when the subjects were divided according to the BMI, this association only remained significant in subjects with BMI < 25 (Figure 2d,e).  16 35 Abbreviations: BMI-Body mass index; CRC-colorectal cancer.

Methylation of SFRP2 in CRC Tissue is Associated with BMI
In order to investigate whether adiposity affects DNA methylation levels, the methylation pattern of SFRP2 and LINE-1 was compared in tumor tissue and tumor-free area, in both BMI groups. Our results showed that the SFRP2 methylation was significantly lower in overweight/obese individuals (34.06%), compared to the non-obese individuals (51.24%) (Figure 2A), while LINE-1methylation did not show any significant difference in BMI (56.68% for non-obese subjects and for overweight/obese individuals 57.69%) ( Figure 2B). We also observed a significant negative correlation between global LINE-1 and SFRP2 methylation in CRC tumor tissue (r = −0.329, p = 0.011) ( Figure 2C). However, when the subjects were divided according to the BMI, this association only remained significant in subjects with BMI < 25 ( Figure 2D and 2E).

SFRP2Methylation in CRC Patients with and without Neoadjuvant Treatment
The specific location of the tumor is a factor to be considered, as they could present biological and clinical differences. We analyzed the methylation SFRP2 status according to the tumor location. First of all, we observed that SFRP2 methylation was significantly higher on the right-sided CRC (57.33%), as compared to the left-sided CRC (36.60%). However, in this classification, most patient tumor samples were derived from the rectum.

SFRP2Methylation in CRC Patients with and without Neoadjuvant Treatment
The specific location of the tumor is a factor to be considered, as they could present biological and clinical differences. We analyzed the methylation SFRP2 status according to the tumor location. First of all, we observed that SFRP2 methylation was significantly higher on the right-sided CRC (57.33%), as compared to the left-sided CRC (36.60%). However, in this classification, most patient tumor samples were derived from the rectum.
Primary rectal cancer requires specific surgical treatment (total mesorectal excision, preceded by neoadjuvant radiotherapy or chemoradiotherapy); so we aimed to evaluate whether the neoadjuvant therapy could be affecting the methylation levels. Patients who received treatment displayed lower SFRP2 methylation levels than those without treatment, within the tumor tissue, though not in the tumor-free area (Figure 3a). Finally, we assessed the SFRP2 methylation status according to the BMI, in the treated and the non-treated patients. In the group of patients who received neoadjuvant therapy, there was no statistical difference in the SFRP2 methylation levels regarding BMI. However, within the non-treated patients, those with BMI > 25 showed significantly lower SFRP2 methylation levels, when compared to the BMI < 25 group (Figure 3b), albeit only in the tumor samples.
At last, to evaluate whether this association between SFRP2 methylation and therapy could be due to other potential confounder variables, we performed a multivariable analysis, including age, BMI and presence of diabetes mellitus. We proved that neoadjuvant treatment was the main variable explaining the variability of SFRP2 methylation levels (Table 3).
by neoadjuvant radiotherapy or chemoradiotherapy); so we aimed to evaluate whether the neoadjuvant therapy could be affecting the methylation levels. Patients who received treatment displayed lower SFRP2 methylation levels than those without treatment, within the tumor tissue, though not in the tumor-free area ( Figure 3A). Finally, we assessed the SFRP2 methylation status according to the BMI, in the treated and the non-treated patients. In the group of patients who received neoadjuvant therapy, there was no statistical difference in the SFRP2 methylation levels regarding BMI. However, within the non-treated patients, those with BMI > 25 showed significantly lower SFRP2methylation levels, when compared to the BMI < 25 group (Figure 3B), albeit only in the tumor samples.
At last, to evaluate whether this association between SFRP2 methylation and therapy could be due to other potential confounder variables, we performed a multivariable analysis, including age, BMI and presence of diabetes mellitus. We proved that neoadjuvant treatment was the main variable explaining the variability of SFRP2 methylation levels (Table 3).

Discussion
Aberrant DNA methylation in gene promoters has been strongly associated with key dysregulations in the oncogenesis for most human tumor models, including colorectal cancer [29]. Presently, there is a large list of established genes subjected to abnormal DNA methylation that are consequently implicated in pivotal physiological processes, including tumor suppression, which can ultimately prove to be of essential utility as novel, epigenetics-based tumor biomarkers [30].
In this study, we evaluated the methylation status of SFRP2, which functionally acts as an inhibitor of Wnt signaling in CRC, together with BMI influences on such a methylation status. CRC

Discussion
Aberrant DNA methylation in gene promoters has been strongly associated with key dysregulations in the oncogenesis for most human tumor models, including colorectal cancer [29]. Presently, there is a large list of established genes subjected to abnormal DNA methylation that are consequently implicated in pivotal physiological processes, including tumor suppression, which can ultimately prove to be of essential utility as novel, epigenetics-based tumor biomarkers [30].
In this study, we evaluated the methylation status of SFRP2, which functionally acts as an inhibitor of Wnt signaling in CRC, together with BMI influences on such a methylation status. CRC is most commonly initiated by accumulation of β catenin in the Wnt signaling pathway, leading to the activation of the Wnt target genes [31].
However, contradicting results were also published in the scientific literature, stating that the modification of the prognostic impact of obesity by a positive status for nuclear β catenin was associated with a significantly improved survival rate for colorectal cancer [32].
The results of our study demonstrated that decreased SFRP2 methylation in CRC was associated with an increased BMI. It has been established that adipose tissue is associated with local inflammation due to increased production of cytokines and pro-inflammatory factors through the macrophages present in adipose tissue [33]. This inflamed state was able to create a chronic low-grade inflammation that can activate the Wnt/β catenin pathway, where SFRP2 plays an important regulatory role in the carcinogenesis process of CRC [34]. In essence, the functional role of SFRP2 in CRC and its potential influence through an increased BMI, could explain, at least partially, the process through which obesity and Wnt/β catenin might exert their effect on carcinogenesis through independent molecular pathways [35].
The SFRP2 methylation status was found to be associated with a decreased gene expression of SFRP2 [36] and, furthermore, was found to be involved in carcinogenesis through down-regulation of tumor suppressor genes. Indeed, SFRP2 can be detected in many biological samples, such as serum and stools, rendering it a potential non-invasive biomarker for many carcinogenic processes [37]. In this study we discovered that SFRP2 was hypermethylated in CRC tumor tissue, when compared to adjacent tumor-free tissue areas. In addition, the hypermethylated SFRP2 in vitro (as found in the HTC116 CRC tumoral cell line) were presumably associated with suppression of gene expression. Following treatment with AZA as a demethylating agent, the expression of SFRP2 was significantly increased and restored. This suggests that SFRP2 demethylation has a functional role in the transcriptional control of SFRP2. Our study was not focused on exploring the potential mechanisms involved in this decrease of DNA methylation pattern, although a possible explanation could be due to a de-regulation of DNA methyltransferase 1 (DNMT1). In most tumors, an altered regulation of this enzyme was described, which is involved in maintaining methylation patterns [38]. Studies have shown that DNMT inhibition promotes gene expression of SFRP2, suggesting DNMTs can modulate SFRP2 methylation status [39].
Genome-wide DNA hypo-methylation of LINE-1 is associated with a worse prognosis in early-stage colorectal cancer, though the majority of the performed studies did not reveal a relation between LINE-1 hypomethylation and clinical outcome in more advanced disease stages [40]. However, our study results confirmed this association, since we revealed a low methylation rate for LINE-1 in tumor samples, without any association or relationship with the clinical-pathological parameters of colorectal tumors.
Previous studies have reported that the location of tumors could influence treatment choice, whereby a large percentage of survival patients were specifically patients who presented primary CRC tumors on the left-side of the abdomen [41]. Consequently, we focused our study on SFRP2 methylation in rectum tumor samples. Treated patients demonstrated a decrease of SFRP2 methylation, independent of the BMI value. This supports the hypothesis that both SFRP2 methylation and BMI could have a potential role in the development of CRC. Additionally, in the non-treated group, we observed higher SFRP2 methylation levels in patients with low BMI, suggesting a possible tumor suppressor role of SFRP2 in overweight/obese patients.
These results point to a possible interaction between BMI and neoadjuvant treatment. Moreover, recently, described epigenetic changes in DNA have also been underlined as mediators of the networking between obesity and cancer [42,43]. Li et al. reported that obesity led to epigenetic remodeling that resulted in differential gene expression related to metabolism and tumorigenesis. Moreover, this epigenetic signature could be reverted by weight loss reinforcing the role of BMI in the epigenetic mark of genes involved in neoplasia [44].
Limitations of this research include the lack of large cohort groups in order to analyze the effect of therapy with more statistical significance and perform survival studies. We could also not demonstrate the causal relation between methylation and gene expression, since expression data were obtained from human colon cancer cell lines. Finally, due to the challenge involved in obtaining high integrity RNA from tumor paraffin-embedded samples, we could not evaluate the gene expression from different DNA methyltransferases.

Conclusions
We demonstrated that the DNA methylation levels of SFRP2 are lower in overweight/obese patients, in comparison with non-obese patients, specifically in tumor areas, supporting the hypothesis that BMI could play a protective role in the development of CRC. We also examined that the patients undergoing neoadjuvant treatment showed similar methylation values of SFRP2, independent of their BMI, suggesting that neoadjuvant treatment might have an effect on the epigenetic regulation of SFRP2, though further studies of clinical pathologies for such patients would be necessary to strengthen these preliminary results.