October 15, 2014 (Vol. 34, No. 18)
Ken Doyle, Ph.D.
Complex diseases such as cancer involve multiple cellular pathways that are affected by the regulation of multiple genes.
The rapid development of technologies to study gene expression has enabled new approaches to studying many of these pathways, generating a volume of information that was not thought possible just a decade ago.
Most of these technologies initially focused on the difference in levels of mRNA expression between tissues in the normal and diseased states. Increasingly, the scope of cancer research has expanded to include the regulation of gene expression by microRNAs (miRNAs) and to study how epigenetic modifications affect gene expression.
In addition to its normal roles from early embryonic development to aging and senescence, DNA methylation is implicated in various pathological processes. Jian-Hua Luo, M.D., Ph.D., is the director of the High Throughput Genome Center, department of pathology at the University of Pittsburgh School of Medicine. In a recent study, Dr. Luo and collaborators examined how DNA hypermethylation may contribute to the development of prostate cancer.
Dr. Luo notes that DNA methylation analysis poses a unique set of challenges. “The first challenge,” he says, “is the cost of whole-genome sequencing. At present, it is over $3,000 per sample.” Additional challenges, according to Dr. Luo, include the time required for analysis, and the difficulty of interpreting the results.
The study used bisulfite sequencing on an Illumina HiSeq 2000 system to examine differential methylation patterns between tumor and normal samples. These results were supplemented with RNA-seq data, and the researchers employed five different analytical techniques to perform concordance analysis between the DNA methylation and gene expression results.
“We found tremendous differences in methylation patterns between healthy organ donor prostates that are age-matched with the prostate cancer patients,” informs Dr. Luo. “However, when we compared CpG island methylations between prostate cancer and matched benign prostate tissues adjacent to cancer, such differences largely disappeared.” Dr. Luo explains that cancerous epigenetic alterations likely occurred in morphologically benign prostate tissues and potentially preceded the development of cancer.
“Promoter CpG island methylation is not as universal and effective in suppressing gene expression as we thought,” remarks Dr. Luo. Since methylated CpG islands that correlate with suppressed gene expression are enriched in Sp1 sites, he speculates that the suppression could result from a cooperative interaction between enhancer-binding proteins and a methylcytosine-binding protein.
DNA Methylation, miRNAs, and Breast Cancer
Altered patterns of DNA methylation are also implicated in various breast cancers. These changes include both global hypomethylation and gene-specific hypermethylation. William B. Coleman, Ph.D., professor in the department of pathology and laboratory medicine at the University of North Carolina School of Medicine, has extensively studied these associations. Dr. Coleman’s recent research focused on the gene encoding DNA methylation enzyme DNMT3b, which is post-transcriptionally regulated by miRNAs.
The study examined miRNA expression in paraffin-embedded primary breast tumors and normal mammoplasty tissue samples, using qPCR. “Specific miRNAs were selected for evaluation based upon known regulators of DNMT3b mRNA,” explains Dr. Coleman.
“We found gene promoter hypermethylation among genes that are silenced or display reduced expression in these cell lines and tissues, with some genes commonly silenced and others variably silenced,” he says. In addition, the study observed that some DNA methylation events occurred commonly and others occurred uniquely across the cell lines and tissues studied. miRNA expression was not directly correlated to DNA methylation or gene silencing events; however, it correlated with the overall DNA methylation status.
“When the expression of regulatory miRNAs was compromised, the number of DNA hypermethylation-associated gene silencing events increased,” observes Dr. Coleman. This observation is consistent with the loss of regulation of DNMT3b expression and consequent DNMT3b enzymatic hyperactivity.
Dr. Coleman explains that DNMT3b dysregulation and aberrant DNA hypermethylation occurs predominantly—but not exclusively—in the triple-negative breast cancer subtypes. Some triple-negative breast cancers do not display aberrant DNA hypermethylation. Therefore, he concludes, miRNA dysregulation is a later molecular event that contributes to the phenotypic characteristics of a subset of triple-negative (basal-like and claudin-low) breast cancers.
DNA Methylation and Ovarian Cancer
Another study examined the relationship between genome-wide DNA methylation in leukocytes and ovarian cancer. “In the United States, ovarian cancer is the 11th most common cancer but the 5th leading cause of cancer deaths,” says Ellen Goode, Ph.D., a researcher and genetic epidemiologist at the Mayo Clinic.
Dr. Goode reiterates the technological challenges associated with whole-genome methylation analysis. In particular, sampling blood creates its own set of problems. Since buffy coat DNA is derived from a mixture of white blood cells, with each type having unique methylation patterns, this can complicate methylation analysis.
The study built on the results of an earlier report that also analyzed DNA methylation in peripheral blood in women with and without ovarian cancer. “Unlike that project,” notes Dr. Goode, “we also examined DNA methylation in relation to the overall survival of ovarian cancer cases.” Her study used a relatively large sample size, excluded CpGs known to associate with white blood cell type, was restricted to pretreatment cases, and used a denser 450K Illumina methylation array.
“We did not find any CpGs that were strongly differentially methylated among cases with respect to survival, but several loci showed highly significant differential methylation when we compared ovarian cancer cases to our matched controls,” explains Dr. Goode. These CpGs were often located near genes that were also highlighted in the previous study, including HDAC3, a histone deacetylase involved in cell cycle progression and PAG1, which encodes a transmembrane adapter protein implicated in T-cell activation.
“This field is just opening up,” concludes Dr. Goode. “We are working actively to uncover relationships between inherited variation, lifestyle factors, and DNA methylation in ovarian cancer.”
Cell Membrane Protein Glycosylation
Post-transcriptional modification of cell membrane proteins is another area of intense focus in cancer research. Nicolle Packer, Ph.D., is a professor of functional proteomics at Macquarie University who studies how changes in glycosylation of membrane proteins are associated with the development of ovarian cancers. Dr. Packer and her colleague Merrina Anugraham note that patients diagnosed with ovarian cancer typically face a poor prognosis for several reasons. “There is a lack of distinctive symptoms and specific tumor markers for early-stage detection,” says Dr. Packer. Chemotherapy resistance is common in advanced stages of the disease, adds Anugraham.
Their study focused on a class of cell membrane protein modifications known as N-glycans, which play a variety of roles in cell adhesion and signaling. These sugars are often associated with tumor invasion and metastasis.
The study used a combination of technologies. First, the N-glycans released from proteins were analyzed by porous graphitized carbon (PGC)-liquid chromatography mass spectrometry. This technique, says Anugraham, “has the amazing capacity to differentiate between isomers of the sugars attached to the protein that have the same composition of monosaccharides, but different structures that present different epitopes on the surface of the cells.”
Next, the researchers used qRT-PCR to correlate the glycan structural changes with the enzymes and genes responsible for producing these differing structures. To investigate potential epigenetic influences on glycosyltransferase enzyme gene regulation, the study used the DNA methyltransferase inhibitor 5-aza-2´-deoxycytidine.
“Ovarian cancer cells express several forms of unique glycan structures,” says Dr. Packer. “These include bisecting N-acetyl-glucosamine (GlcNAc), LacdiNAc, and α2-6 sialylated types of N-glycans.” There was, Anugraham adds, a corresponding increased expression of specific glycosyltransferase genes—MGAT3, B4GALNT3, and ST6GAL1—in ovarian cancer cells. This result correlated with the increased expression of these membrane glycan types observed by mass spectrometry.
Dr. Packer concludes, “Our results could potentially be used for the development of novel antiglycan drug targets and may facilitate the discovery of new tumor markers for the early detection of ovarian cancer.”
In addition to employing an array of technologies to study gene expression, researchers are developing sophisticated analytical methods to decipher the vast amount of gene expression data already in repositories. One such study analyzed data from patients with glioblastoma multiforme (GBM), an aggressive form of brain cancer that accounts for approximately 70% of all malignant gliomas.
Karl T. Kelsey, Ph.D., is the director of the Center for Environmental Health and Technology, Brown University. “Our study uniquely focused on methylation and attempted to classify the effects of methylation on survival into those mediated by expression and those not mediated by expression,” he says. The study analyzed data from The Cancer Genome Atlas (TCGA) to examine methylation-directed changes in gene expression as well as alternative associations of DNA methylation with disease survival.
The study observed significant associations in 27 unique methylation/expression pairs. “The majority of the predictive DNA methylation loci were located within CpG islands, and all but three of the locus pairs were negatively correlated with survival,” notes Dr. Kelsey. These observations suggest that, for most loci, methylation/expression pairs are inversely related, consistent with the regulatory role of methylation in GBM. Dr. Kelsey adds that the approach may provide a method for biomarker discovery to identify novel sites for locus-specific drug targets.