Methylomics of breast cancer: Seeking epimarkers in peripheral blood of young subjects

Abstract

Critical roles of epigenomic alterations in the pathogenesis of breast cancer have recently seized great attentions toward finding epimarkers in either non-invasive or semi-non-invasive samples as well as peripheral blood. In this way, methylated DNA immunoprecipitation microarray (MeDIP-chip) was performed on DNA samples isolated from white blood cells of 30 breast cancer patients compared to 30 healthy controls. A total of 1799 differentially methylated regions were identified including SLC6A3, Rab40C, ZNF584, and FOXD3 whose significant methylation differences were confirmed in breast cancer patients through quantitative real-time polymerase chain reaction. Hypermethylation of APC, HDAC1, and GSK1 genes has been previously reported in more than one study on tissue samples of breast cancer. Methylation of those aforementioned genes in white blood cells of our young patients not only relies on their importance in breast cancer pathogenesis but also may highlight their potential as early epimarkers that makes further assessments necessary in large cohort studies.

Background

Lifestyle and environmental factors as well as genetic factors play important roles in breast cancer development.¹ Epigenetic alterations which may be induced by external exposures could modify and increase the genetic susceptibility among breast cancer patients who are negative for well-known genetic mutations.² Finding the exact epigenetic mechanism in pathogenesis of breast cancer, therefore, could pave the way toward early screening and diagnosis and even targeted treatment of affected patients.

Epigenetic inter-individual variations namely methylation changes have recently highlighted in Cancer Genetics that may lead us to recover the black-boxes of record successive events during development of different disease as well as breast cancer.³ DNA methylation as inheritable epigenetic mechanisms could be seen in promoter or intragenic regions with different effects on gene expression. The role of methylation changes in tumor development, independent of their influences on gene expression levels, has been addressed in several cancer studies.^4,5

Moreover, it seems that methylation pattern may predict cancer development in pre-malignant tissues.^6,7 However, biopsy of targeted tissue as well as breast is not feasible in the screening programs. Peripheral blood can be served as a semi-non-invasive source of methylation analysis which could in turn suggest epimarkers that may potentially be linked to the development of breast cancer. The exact correlation between the methylome of primary cancer tissue and white blood cells (WBCs) remained to be elucidated.⁸ There are emerging evidences, nonetheless, which have demonstrated the potential of WBC methylation pattern in prediction of different cancers such as pancreatic, ovarian, bladder, and breast.^9–14 In this regard, we previously reviewed the epigenomic map of WBCs, which can be employed as a potential biomarker in breast cancer.¹⁵ Some prospective studies indicated that methylation changes could be detected in WBC genome even years in advance of cancer development, as an alarm for cancer predisposition.^16–18 In this regard, to ignore the possible bias of the effect of aging on genomic methylation pattern, investigating the methylation changes in young women seems to be more satisfactory. In addition, owing to the increasing rate of breast cancer in young women with a more aggressive and poor prognosis pattern, early identification of these patients through non-invasive WBCs methylation analysis would be a promising road toward alleviating the various economical and psychological burdens in young societies.¹⁹

In this study, we were aimed to primarily seek the methylome of peripheral blood samples in young breast cancer patients compared to healthy women through a whole genome approach called methylated DNA immunoprecipitation microarray (MeDIP-chip).

Material and methods

Samples preparation

Thirty breast cancer patients who had no familial history of breast cancer and underwent breast-saving surgery or mastectomy were recruited from Imam Khomeini Hospital. Absolutely normal samples without any familial history of cancer were selected to be as healthy controls. They were selected from the same ethnicity as well as cases and were referred to do either screening mammography or breast sonography. All the enrolled samples including cases and controls were aged 25–35 years (mean age in both groups: 30 ± 0.40) and filled the informed consent form according to the ethical committee of Tehran University of Medical Sciences. Blood samples collected after any intervention including surgery, chemotherapy or radiotherapy were excluded from the study.

DNA methylation analysis

DNA was extracted from whole blood using the High Pure PCR Preparation Kit (Roche), and then, it was sonicated to obtain size range of fragment (200–1000 bp) optimal for MeDIP analysis.

Sonicated DNA was subjected to MeDIP process in which all the methylated DNA sequences are pulled down by a monoclonal antibody against 5-methylcytosine. MeDIP was performed using the Methylated DNA Capture Kit (Epigentek) according to the manufacturer’s instructions. Enriched methylated DNA sequences were amplified in quantitative real-time polymerase chain reaction (qPCR) using primers specific to H19 gene as the endogenous positive control (hemimethylated gene) and GAPDH as the unmethylated gene. The same amplification reaction was done on input DNA samples which were the fraction of sonicated DNA including both methylated and unmethylated sequences that have been left before MeDIP analysis to be used as the reference DNA control. Enrichment was calculated using the following equation²⁰

$$\mathrm{Enrichment}=\frac{{({\mathrm{E}}_{\mathrm{target}})}^{\Delta \mathrm{Cq}\mathrm{target}(\mathrm{Input}-\mathrm{MeDIP})}}{{({\mathrm{E}}_{\mathrm{reference}})}^{\Delta \mathrm{Cq}\mathrm{reference}(\mathrm{Input}-\mathrm{MeDIP})}}$$

where “E” is the efficiency of amplification, “target” is the methylated region of interest, and “reference” is the normalizer (a DNA region which is known, or expected, to be unmethylated in all the samples). The enrichment ranges of 2⁵–2⁷ were included for the following steps of analysis.

Since the concentration of immunoprecipitated (IP) samples was too little to be analyzed in microarray, the GenomePlex Complete Whole Genome Amplification (WGA) Kit (#WGA2; Sigma-Aldrich) was used to amplify the IP and input fractions obtained from all of the case and control samples.

The input and IP DNA fractions were labeled with Cy3 and Cy5 dyes, respectively, and then, they were hybridized to the NimbleGen Human DNA Methylation 3×720K Promoter Plus CpG Island Array. It is a multiplex slide with three identical arrays per slide in which each array contains 27,728 CpG islands annotated by UCSC and 22,532 well-characterized RefSeq promoter regions (from about −2440 to +610 bp of the transcription start sites (TSSs)) have been totally covered by ~720,000 probes.

Statistical analysis

Scanning was accomplished with the Axon GenePix 4000B microarray scanner. Raw data created by NimbleScan software were normalized using median-centering, quantile normalization, and linear smoothing by Bioconductor packages Ringo, limma, and MEDME. After normalization, normalized log₂-ratio data were constructed for each sample. Enriched peaks should be generated from normalized log₂-ratio data.

Each probe obtained a −log₁₀ p-value score from the windowed Kolmogorov–Smirnov (KS) test. If several adjacent probes rose significantly above a set threshold, the region was assigned to an enrichment peak (EP). NimbleScan detected the peaks by searching for at least two probes above a p-value minimum cutoff (−log₁₀) of 2.

The average of log₂-ratio value (Experiment and Control) was used to calculate the M′ value (defined by the following equation) to compare the probe-specific differentially enriched regions between two studied groups. The NimbleScan sliding-window peak-finding algorithm was then rerun on those data to find the differential enrichment peaks (DEPs)

$$\mathrm{M}\prime =\mathrm{Average}\left(\frac{\mathrm{lo}{\mathrm{g}}_2\mathrm{MeDIPE}}{\mathrm{InputE}}\right)-\mathrm{Average}\left(\frac{\mathrm{lo}{\mathrm{g}}_2\mathrm{MeDIPC}}{\mathrm{InputC}}\right)$$

The DEPs determined by the NimbleScan algorithm were filtered according to the following criteria:

• SLC6A3 (chr5: 1469792–1470004 HG18) which was hypermethylated in CpG island of the fifth intron.

• Rab40C (chr16: 604037–604588 HG18) which was hypermethylated in CpG island of the first intron.

• ZNF584 (chr19: 63604085–63604338 HG18) which was hypomethylated in promoter region.

• FOXD3 (chr1: 63604338–63562256 HG18) which was hypermethylated in promoter region.

Real-time PCR validation

Real-time PCR was performed to confirm the methylation difference of SLC6A3, Rab40C, ZNF584, FOXD3, TP53, HPCAL1, MPRS, and NEUROG2 genes between cases and controls. The precision and variability of the test was calculated based on the Ct variation from the Ct mean value.

The CV value of SLC6A3, Rab40c, ZNF584, and FOXD3 genes in the case group was low (<5%), and also, mean value comparison in the mentioned genes was significant between cases and controls (p < 0.05). It seems that the RQ value in the aforementioned genes is considerable between two case and control groups (Table 2).

Table 2.

The mean of ΔCT and CV values between breast cancer patients and control samples.

Gene	Breast cancer cases		Healthy controls		p*
	Mean ΔCT (n = 30)	CV (%)	Mean ΔCT (n = 30)	CV (%)
SLC6A3	23.02	0.06	25.01	0.08	0.031
RAB40C	25.09	0.09	26.66	1.07	0.022
ZNF584	24.01	1.73	22.08	1.01	0.024
FOXD3	26.06	1.32	27.12	0.82	0.012
TP53	23.07	7.9	25.03	1.04	0.076
HPCAL1	24.05	8.3	26.04	1.08	0.081
MPRS	24.01	9.2	23.07	0.2	0.083
NEUROG2	23.04	7.4	21.03	7.4	0.072

CV: coefficient of variability.

*

p values less than 0.05 were considered as significant.

Finally, obtained results demonstrated that the methylation pattern of SLC6A3, Rab40c, ZNF584, and FOXD3 genes in the leukocytes of peripheral blood could be used as a potential biomarker for the risk assessment of breast cancer.

Discussion

Germ line mutations in BRCA1/2 genes have been described in less than 5% of all breast cancers including early-onset diseases.^21,22 Shedding light on the role of other genetic and epigenetic susceptibilities in BRCA1/2 mutation–negative early breast cancer patients may open the new avenues toward their successful targeted diagnosis and treatment. In this study, 30 breast cancer patients comparing to healthy control samples were analyzed to determine their peripheral blood methylome as predictive epimarkers. To the best of our knowledge, this is the primary peripheral methylome analysis focused on young sporadic breast cancer patients aged 25–35 years through a whole genome approach.

Two novel DMRs located in intragenic regions of SLC6A3 and Rab40c genes and two DMRs in the promoter of ZNF584 and FOXD3 genes were identified in breast cancer patients.

SLC6A3 gene encodes a dopamine transporter belonged to a sodium- and chloride-dependent neurotransmitter transporter family, and it is expressed in multiple tissues as well as WBCs. The CpG island localized in the intron 5 of this gene was shown to be significantly hypermethylated in cancer patients compared to healthy controls. SLC6A3 gene consists of 27 CpG islands in which 21 islands are localized within the intronic regions, signifying the important role of those CpG islands in regulation of gene expression.²³ SLC6A3 was previously enlisted as genes which have been methylated within promoter and intragenic regions in the tissue samples of triple-negative breast cancer patients and was shown to be associated with poor prognosis.²⁴ Peripheral promoter hypermethylation of SLC6A3 was demonstrated in chronic physical aggression and schizophrenia.^25,26

The second DMR was Rab40c which is a member of the RAS oncogene family.²⁷ Herein, the CpG island located in the first intron of this gene was primarily found to be hypermethylated in breast cancer patients. In contrast, Rab40c was identified as one of the hypomethylated gene-body DMRs in TMPRSS2:ERG fusion–positive prostate cancer patients which was not associated with a high level of gene expression.²⁸ There are paradox investigations indicating the over-expression of some Rab family members^29–35 in comparison with hypermethylation and low expression of some other Rab family members in various types of cancers.^36–38 Although the tumor suppressor role of Rab25 family member has been identified in a luminal subtype of breast cancer,³⁹ it is thought that the exact role of Rab family proteins as oncogene or tumor suppressor genes is dependent on the accessibility of them to modifiers.⁴⁰ The other scenario says that some microRNAs that are involved in cancers influence the tumorigenesis process through targeting of the Rab family member,^41–43 that is, microRNA-200b which is implicated in breast cancer as a prognostic factor since it was demonstrated to affect some of Rabs.⁴⁴ Rab40c gene expression was shown to be dramatically increased during two phases of inflammation and migration–proliferation.⁴⁵

Intragenic methylation in contrast to promoter methylation is not always associated with gene silencing and may be detected in normal tissues. It was suggested that intragenic CGI could be the site of alternative promoter or initiation of non-coding RNA (nc-RNA) transcription with prominent regulatory effects on gene expression.⁴⁶ Therefore, hypermethylation of intronic CpG islands within the Rab40c and SLC6A3 genes is supposed to interfere with transcription of nc-RNA and therefore leads to deregulation of gene expression. Further studies are required to elucidate the functional values of intragenic regions and the role of their epigenetic patterns in regulation of gene expression.

ZNF584 promoter was found to be significantly hypomethylated in our patients compared to healthy women. ZNF584 is one of the KRAB-ZNF transcriptional regulatory family proteins that regardless of their great quantity, little is known about their gene targets and physiological functions.^47,48 They modulate crucial cellular procedures such as proliferation, apoptosis, and neoplastic transformation. Their functions are dependent on cell-specific factors that recruit them for transcriptional management, and therefore, they could play as oncogenes or tumor suppressors during the development of cancer.⁴⁹ A concomitant increase in the copy number of ZNF584 gene in breast cancer tissues and blood samples found in a recent study⁵⁰ could be consistent with promoter hypomethylation detected in this study. In addition, among ZNF family members, ZNF217 was demonstrated as a DMR in WBCs of breast cancer patients⁴ besides the other member, ZNF584, which was identified as a hypomethylated gene in the peripheral blood of these patients.

FOXD3 is another DMR that was significantly hypermethylated in breast cancer patients. It is belonged to the forkhead family of transcription factors which are characterized by a distinct forkhead domain.⁵¹

FOXD3 has been characterized as a tumor suppressor gene in multiple cancers, and its reduced expression may be due to promoter hypermethylation as a major mechanism of gene silencing.^52,53 It was demonstrated that promoter methylation associated with lower expression of some stem cell determinant genes including FOXD3 induced differentiation of stem cells toward malignancy.⁵⁴ The pivotal role of FOXD3 expression in suppression of tumorigenesis was confirmed in two metastatic breast and colorectal cancers wherein the promoter methylation disabled its expression and inhibitory function on B-Raf-induced cell proliferation.^51,55 Methylation of FOXD3 gene was previously detected in Helicobacter pylori–induced gastric cancer cells⁵⁶ as well as in chronic lymphocytic leukemia (CLL).⁵⁷ Decreased FOXD3 gene expression was described to be relevant with epithelial–mesenchymal transition (EMT) in breast cancer.⁵⁸ Hypermethylation of FOXD3 gene in the peripheral blood of our early breast cancer patients which is consistent with previous cancer studies may be relying on the critical role of its expression in prevention of inactivation of downstream tumor suppressor gene in early stages of cancer.

It is interesting to note that CpG-rich genes including those mentioned above are conserved against methylation, and detecting the methylation of them in breast cancer is a promising issue that should be considered in epimarker researches.

Although there are some reports on methylome analysis of breast cancer, most of them have focused on breast cancer tissues or cell lines. To our knowledge, this is the third genome-wide methylome analysis performed on peripheral WBCs of breast cancer patients. Two other previous studies were carried out on discordant twin pairs affected by different cancers including breast.^18,59 Neither of DMRs identified in the recent study was found in our methylome analysis.⁵⁹ However, TP53 promoter methylation found in the peripheral blood of our young breast cancer patients may be the preceding event to methylation of its downstream gene, COX7C, which was detected to be methylated in a previous study.⁵⁹ In the second prior study on WBCs methylome, hypermethylation of PDE4C, MPPE1, and PRR5 genes was found to be in common with our findings, highlighting the possible importance of their expression in prevention of breast cancer development.¹⁸ There are some other candidate genes methylation analyses on WBC samples of breast cancer patients^9,16,60–66 in which DMRs in the BRCA1, RAD51C, ATM, and NEUROD1 genes were found to be similar with our result.

In the other candidate methylation study on breast cancer, RASSF1A, APC, HIN1, BRCA1, cyclinD1, RARβ, CDH1, and TWIST1 genes were found to be hypermethylated in tumor tissue while not in WBCs. Replication of TWIST1, APC, and BRCA1 genes hypermethylation in WBCs of our young samples may indicate their critical inhibitory role in breast carcinogenesis process and their potentials as early peripheral biomarkers.⁶⁷ Moreover, hypermethylation of peripheral WBCs genes including ESR1, GSTM2, MAGEA1, MSI1, SIX3, EN1, PAX3, GSK1 and APC in the present study have previously been found in a targeted methylation analysis of breast cancer tissues insisting on the importance of APC gene methylation further again may be as an early event.⁶⁸ Among the aforementioned gene list, GSK1 were also demonstrated to be hypermethylated in familial breast cancer tissues accompanying PKD2, CXCL1, CDCP1, PB1, and GADD45 genes which is consistent with our methylome findings on WBC samples.⁶⁹ In the other methylation profile found in breast tissue samples obtained from infiltrative ductal carcinoma, hypermethylation of SOX1, FGF2, SPARC, SOX17, and TAL1 genes was found to be in common with our results.⁷⁰

The consistent methylation pattern between tumor tissue and peripheral WBCs in the same ethnicity samples is an inevitable characteristic of an epimarker before it is widely validated and approved.⁷¹ Association between WBC methylation and solid tumor has been explained in some mechanistic level including constitutional epimutation, epigenetic changes due to genetic variation and/or environmental exposures, and finally immune system responses due to cancer inflammation or cancer-associated environmental exposures.¹⁵ It is worth to note that the significant difference in methylation status of genes found in the peripheral blood of breast cancer patients may be affected by the change in population of different types of WBCs as previously described by the effects of tobacco on peripheral blood hypomethylation.⁷²

Nonetheless, we found some hypermethylated genes in WBCs of breast cancer patients including APC, HDAC1, BRCA1, and GSK1 whose hypermethylation pattern has been previously reported in more than one studies of breast cancer tissues. Replication of the same gene methylation pattern in different sample types and patients’ race may highlight the critical epigenetic changes of those genes toward their under-expression leading to cancer progression. The panel of methylated genes in this study would be considerable from two points of view. First, it is to be noted that breast cancer patients of age 25–35 years were selected, which is our major different approach in epigenetic analysis though the age range of other breast cancer methylome analysis was often more and sometimes neglected to be published. Second, methylome of the young patients was assayed on WBCs as an emerging source of methylation analysis for semi-non-invasive cancer diagnosis and follow-up.

Aging is a powerful risk factor for cancer-associated gene aberrations due to the fact that different exposures in the life span may affect the methylation pattern of genes involved in cancer development. In this respect, some studies demonstrated that age-associated methylation occurs in cancer-related genes including polycomb group targets and age-associated DMRs observed in peripheral DNA, exhibiting a methylation profile of various germ cell layers. One of the key features of this study was selection of young sporadic breast cancer patients to avoid age-induced gene methylation. Collectively, epigenetic signature of peripheral blood can be a surrogate of external exposure as well as genetic variability.⁶⁹ However, alteration of methylation status of specific genes may reveal the new targets of the second hit sparking at the inherited harmless first hit to affect young women with breast cancer. Further studies are warranted to determine the methylation status of the presented DMRs in tissues and WBCs of breast cancer patients in the large cohort studies and in various populations with racial disparities.

This article was extracted from a part of the PhD thesis of Dr Golnaz Khakpour, supervised by Dr Javad Tavakkoly-Bazzaz and Dr Mehrdad Noruzinia.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This research was financially supported by Tehran University of Medical Sciences, Tehran, Iran (grant number: 18966).