Histone modifications and cancer: biomarkers of prognosis?

Abstract

Epigenetic dysregulation is being increasingly recognized as a hallmark of cancer. Post-translational modifications of histones, in particular, are known to play important roles gene expression alterations in cancer development and progression. Given their key involvement in the various stages of carcinogenesis, histone modifications are also being explored as potential biomarkers of disease progression and prognosis. This review will therefore discuss the role of histone modifications in cancer biology and will explore their prognostic potential.

Introduction

Cancer has long been acknowledged as a pleiotropic and multifaceted disease, the initiation and progression of which, is impacted by a myriad of factors. Given its remarkable intricacy and complexity, the concept of cancer as a disease of epigenetic, as well as genetic, alterations has gained a considerable momentum within the scientific community [1]. Epigenetics is the study of inherited phenotypes, which are not encoded by the DNA sequence [2,3]. With respect to cancer, the term epigenetics commonly refers to changes in DNA methylation, microRNAs, histone post-translational modifications, and other chromatin elements that can alter gene expression [4]. In the 1990s, one of the focal points of epigenetic cancer research was to further illuminate the 1980s discovery of DNA methylation abnormalities [5]. However, during the past decade, this focus has been broadened by the upsurge of research regarding the role of covalent chromatin modifications (i.e. DNA methylation and post-translational modifications) in gene regulation, carcinogenesis, and cancer prognosis [6,7].

Although not as widely studied as DNA methylation, post-translational modification of histone tails, play a critical role in chromatin regulation, gene activity and nuclear architecture [8,9]. The functional consequences of these modifications can bring about structural changes to chromatin and/or serve to include or exclude protein complexes from coming in contact with the DNA, thereby influencing the transcriptional pattern of genes, altering their activity, and plausibly contributing to the emergence and progression of cancer. This review will therefore focus on discussing the role of histone modification in cancer and cancer prognosis.

Histone modifications and cancer

Histones are highly conserved alkaline proteins that can become post-translationally modified at the amino acid residues located on their N- and C- terminal tails. There are four core histones: histone 2 A (H2A), histone 2 B (H2B), histone 3 (H3), and histone 4 (H4), and one linker histone, histone 1 (H1). Approximately 146 base pairs of DNA are wrapped around each histone octamer, which consists of two copies of each of the core histones, in left-handed superhelical turns. H1, which is not included in the nucleosome “bead”, serves as a linker and helps secure DNA that is wound around the nucleosome [8-11].

Histone residues can become methylated, phosphorylated, acetylated, sumolyated, ubiquitinated, and ADP-ribosylated. Unlike the other modifications, the methylation of amino acids, like lysines and arginines, can vary in amount. Lysine residues (K) can either be mono-, di-, or trimethylated, while arginine residues (R) can be mono-methylated and symmetrically or asymmetrically di-methylated. Notably, both acetylation (ac) and the exact methylation (me) status of lysines (i.e. mono-, di-, and tri-methylation) can influence the state of chromatin (i.e. active vs. inactive) and subsequently the transcriptional status of genes (Strahl BD and Aliis CD, The language of covalent histone modifications, Nature 403: 41-45).

Enrichment in acetylation of histone tails is typically associated with transcriptional activation of genes, while the functional consequences of methylation depend on the number of methyl groups, the residue itself, and its location within the histone tail. For example, histone 3 lysine 4 di- and trimethylation (H3K4me2 and H3K4me3) and histone 3 lysine 9 monomethylation (H3K9me1) are associated with open chromatin and active gene expression, while histone 3 lysine 27 di- and trimethylation (H3K27me2 and H3K27me3) and histone 3 lysine 9 di- and trimethylation (H3K9me2 and H3K9me3) are associated with inactive chromatin and repression of gene expression [12]. In addition, some marks, such as histone 3 lysine 4 mono methylation (H3K4me1) and histone 3 lysine 27 acetylation (H3K27ac) are found in the enhancer elements of genes and can influence gene expression even at large distances from the gene. Active enhancers are enriched with H3K27ac, while those that only bear H3K4me1 are poised for activation in response to a stimuli [13].

The addition or removal of post-translational modifications from histone tails is fairly dynamic and is achieved by a number of different histone modifying enzymes. The enzymes involved in so called “writing” and “erasing” these reversible marks include, histone acetyltransferases (HATs), histone deacetylases (HDACs), histone methyltransferases (HMTs), histone demethylases (HDMs), histone ubiquitinating enzymes as well as deubiquitinating enzymes, and can either be specific (i.e. histone methyltransferases and demethylses) or general (i.e. HATs and HDACs) in their ability to recognize and alter the amino acid residues of histone tails [14-16]. It should be noted that the histone demethylases for example were first discovered in 2006 and these enzymes as well as the histone methyltransferases are likely to modify other proteins in addition to histones. Given the fundamental roles of histone modifications in gene regulation and expression, it is not surprising that aberrant patterns of histone marks are found in cancer. Advances in high-throughput sequencing have allowed for genome-wide mapping of chromatin changes that occur during the tumorigenic process [17]. It was reported that cancer cells experienced a loss of histone acetylation and methylation, with the losses occurring predominantly at the acetylated Lys16 and trimethylated Lys20 residues of histone H4. These losses were also associated with the hypomethylation of DNA repetitive sequences, a well-known characteristic of cancer cells [18].

Inappropriate targeting of histone modifying enzymes, such as HDACs, HATs, HMTs and HDMs, is often responsible for the aberrant histone modifications. HDACs, for example, are often found to be over-expressed in prostate and gastric cancers [19,20]. HDAC1 was shown to associate with the tumor suppressor retinoblastoma protein (Rb), and in cooperation with Rb lead to the repression of transcription factor E2F-regulated promoter of the gene encoding the cell-cycle protein cyclin E [21]. While, aberrant formation of fusion proteins through chromosomal translocations of HAT and HAT-related genes (e.g. MOZ, MORF, CBP and p300) was found to occur in leukemia [22]. Dysregulation of histone methyltransferases or demethylases in cancer cells also contributes to aberrant histone modification patterns. Deletion of EZH2, a H3K27 specific methyltransferase, was associated with high frequency of spontaneous T-cell leukemia occurrence, in mice. Furthermore, EZH2 was found to be highly expressed in prostate and breast cancers [23-25]. G9a, a H3K9 specific histone methyltransferase, was found to promote lung cancer invasion and metastasis by silencing the cell adhesion molecule (Ep-CAM) [26]. LSD1, or KDM1A, a FAD-dependent amine oxidase that demethylates lysine 4 of histone 3 (the mono and di form only), was found to be involved in maintaining the undifferentiated, malignant phenotype of neuroblastoma cells [27]. Inhibition of this demethylase led to a reprogramming of the transcriptome of neuroblastoma cells and inhibited neuroblastoma xenograft growth [27].

Histone modification and cancer prognosis

Cancer is a heterogeneous disease, and can often yield varying clinical outcomes for individuals with the same affected tissue (i.e. prostate or breast cancer). Clinical outcome may be assessed as, but is not limited to, risk of tumor recurrence following primary tumor extraction, risk of metastasis, likelihood of survival, and/or degree of response to therapeutic agents [3]. The ability to anticipate the clinical behavior of cancers is essential in determining the most suitable therapeutic interventions [28]. Considering that cancer is so diverse and clinical outcome predictions often vary from patient to patient, a considerable amount of effort is being invested to discover molecular biomarkers that can categorize cancer patients with distinct clinical outcomes to expand prognostic capabilities. These molecular biomarkers can play roles before cancer diagnosis (in risk assessment and screening), at diagnosis (as discussed in the main text) and after diagnosis (in monitoring therapy, selecting additional therapy and detecting recurrence) and can include single nucleotide polymorphisms, chromosomal translocations, gene mutations, expression patterns of groups of genes, methylation status of specific gene promoters, or secreted proteins, as well as, post-translational histone modifications [3,28]. However, even though the number of potential biomarkers in the literature continues to increase, the number of FDA-approved biomarkers per year is actually decreasing due to the fact that many of the biomarkers, past and present, do not meet the extensive statistical clinical criteria [28].

Although post-translational histone modifications have not been officially inaugurated into the hall of clinical cancer biomarkers, numerous studies have demonstrated that histone modifications can predict the prognosis of various cancers (Table 1). The cancers that will be briefly summarized in this section include: lung, prostate, breast, leukemia, kidney, liver, pancreatic, esophageal, and gastric cancers.

Table 1.

Histone modification patterns predict prognosis in multiple cancers.

Study	Cancer type	Histone Modifications
Song et al. 2012	Lung	H3K9ac, H3K9me3, H4K16ac
Barlési et al. 2007	Lung	H3K4me2, H3K9ac
Van Den Broeck et al. 2008	Lung	H4K5ac, H4K8ac, H4K12ac, H4K16ac, H4K20me3
Seligson et al. 2009	Lung	H3K4me2, H3K18ac
Seligson et al. 2005	Prostate	H3K4me2, H3K18ac
Ellinger et al. 2010b	Prostate	H3K4me1, H3K9me2, H3K9me3, H3Ac, H4Ac
Behbahani et al. 2012	Prostate	H4K20me1, H4K20me2
Bianco-Miotto et al. 2010	Prostate	H3K4me2, H3K18ac
Ellinger et al. 2012	Prostate	H3K27me3
Elsheikh et al. 2009	Breast	H3K18ac, H4K12ac, H3K4me2, H4K20me3, H4R3me2, H4K16ac
Leszinski et al. 2012	Breast	H3K9me3, H4K20me3
Müller-Tidow et al. 2010	Leukemia	H3K9me3
Park et al. 2008	Stomach	H3K9me3
Zhang et al. 2009	Stomach	H3K27me3
Tzao et al. 2009	Esophagus	H3K18ac, H4R3me2,H3K27me3
I et al. 2010	Esophagus	H3K18ac, H4R3me2
Ellinger et al. 2010a	Kindey	H3K4me1, H3K4me2, H3K4me3
Rogenhofer et al. 2012a	Kidney	H3K9me1
Rogenhofer et al. 2012b	Kidney	H3K27me1, H3K27me2, H3K27me3
He et al. 2012	Liver	H3K4me3
Cai et al. 2011	Liver	H3K27me3
Manuyakorn et al. 2010	Pancreas	H3K4me2, H3K9me2, H3K18ac

Lung cancer

Cancers of the lung and bronchus have the highest rate of mortality of any cancer in the United States. American Cancer Society estimates that in 2012, lung and bronchial cancer constitute 29% and 26% of all cancer deaths in men and women, respectively [29]. Primary non-small cell lung cancer (NSCLC) is the main cause of malignancy-related mortality in Asian and Western populations [30]. Clinicopathological analyses of NSCLC tissues demonstrated a positive correlation between lower levels of H3K9ac, H3K9me3 and H4K16ac and tumor recurrence. However when patients were further clustered according to histone modification patterns (i.e. acetylation dominant, methylation dominant, co-dominant and modification-negative), the acetylation-dominant group exhibited better survival prognosis, but methylation-dominant and modification-negative status was associated with poor prognosis [30]. Moreover, NSCLC patients with large-cell or squamous cell carcinomas, whose tumor expressed high H3K4me2 levels and stage I patients with adenocarcinomas with lower levels of H3K9ac had increased survival rates [31]. Histone H4 modifications also displayed an aberrant pattern in lung carcinomas, with hyperacetylation of H4K5/H4K8, hypoacetylation of H4K12/H4K16, and loss of H4K20me3. H4K20me3 was frequent in cell carcinoma and was observed in early precursor lesions, where the level of H4K20me3 staining strongly decreased with disease progression [32]. In adenocarcinomas, the decrease of H4K20me3 was less common, but allowed the detection of a subgroup of stage I adenocarcinoma patients with reduced survival [32]. Decreased levels of H3K4me2 and H3K18ac were also associated with poorer survival probabilities in patients with adenocarcinomas [33].

Prostate cancer

Prostate cancer (PCA) displays a heterogeneous clinical behavior, ranging from slow-growing to highly aggressive, and is the second leading cause of male cancer fatalities in the United States [29]. Immunohistochemical (IHC) examination of primary prostate cancer tissues revealed that even though cancer cells displayed heterogeneity in fraction and intensity of staining, two histone modifications, H3K4me2 and K18ac, which are both marks of transcriptional activation, provided non-redundant information about cancer prognosis. Interestingly, lower cellular levels of histone modifications (i.e. decreased percent of cell staining) were associated with poor patient prognosis [34]. Several other histone marks, such as H3K4me1, H3K9me2, H3K9me3, H3Ac, and H4Ac were also found to be significantly reduced in PCA compared to non-malignant prostate tissue, and H3K4me1 was found to be a significant predictor of prostate specific antigen (PSA) recurrence following radical prostatectomy [35]. Furthermore, prostate cancer tumors were also found to have aberrant pattern of H4K20 modifications, with a general hypomethylation of H4K20me1 and H4K20me2 in hormone naive PCA (mPCA) and castration-resistant PCA (CRPC) [36]. H3K18ac and H3K4me2 were also found to be independent predictors of tumor recurrence in another cohort study [37]. However, this study found that elevated global levels of H3K18ac and H3K4me2 were associated with a 1.71-fold (p < 0.0001) and 1.80-fold (p = 0.006) increased risk of tumor recurrence, respectively. While increased levels of both modifications were associated with a 3-fold increased risk of relapse [37]. An increase in H3K27 methylation and over expression of its specific methyltransferase, zeste homologue 2 (EZH2), were also shown to be associated with PCA progression and metastasis [25,38].

Breast cancer

Comparable to prostate cancer, breast cancer is a heterogeneous disease, ranging from premalignant hyperplasia to invasive and metastatic carcinomas, and is the second leading cause of cancer related mortalities in women [29]. Examination of well characterized series of human breast carcinomas revealed that there was a highly significant correlation between global histone modifications status, tumor biomarker phenotype, and clinical outcome. Elevated levels of global histone acetylation and methylation were associated with a favorable prognosis and detected almost exclusively in luminal-like breast tumors. While, moderate to low levels of H3K18ac, and H4K12ac, as well as, H3K4me2, H4K20me3, and H4R3me2, were observed in carcinomas of poorer prognostic subtypes, including basal carcinomas and HER-2-positive tumors. Moreover, this analysis also revealed that low or absent H4K16ac in the majority of breast cancer cases, suggesting that this alteration may represent an early sign of breast cancer [39]. In addition to alterations in the global levels of histone modifications, breast cancer also appears to exhibit gene specific histone alterations. Chromatin-immunoprecipitation (ChIP) analysis of breast cancer tissues revealed that centromeric satellites (SAT2) level were up-regulated with H3K9me3, a mark of transcriptional repression, and with H4K20me3, a mark that also plays a role in gene silencing, as well as, indexing of pericentric heterochromatin [40].

Aberrant expression of a number of histone modifying enzymes has also been correlated with breast cancer prognosis. Over expression on EZH2, which is part of Polycomb Repressive Complex 2 (PRC2) and methylates histone 3 lysine 27 (H3K27), was shown to correlate with breast cancer aggressiveness and poor patient prognosis [24,41]. LSD1, or KDM1A was also found to be highly elevated in estrogen receptor (ER)-negative tumors [42]. Moreover, patients whose tumors had higher expression of histone deacetylase 6 (HDAC6), which is a late responsive estrogen-induced up-regulated gene, had a better prognosis than those with low levels of this enzyme, in term of disease free-survival [43].

Leukemia

The American Cancer Society estimates that 47,150 new cases of leukemia are expected in 2012 [29]. Leukemia is a cancer of the bone marrow and blood and is categorized according to cell type and rate of growth into four main groups: acute lymphocytic (ALL), chronic lymphocytic (CLL), acute myeloid (AML), and chronic myeloid (CML). Almost 90% of leukemia cases are diagnosed in adults, with AML and CLL being the most common [29]. ChIP-ChIP analysis of AML patient samples revealed that hundreds of promoter regions were associated with decreased H3K9me3 levels and the H3K9 modification signature improved prognosis prediction independent of karyotype, age, and NPM1/ FLT3 mutations [44]. ChIP-ChIP anlysis from another large AML patient cohort also revealed that dyslocalizatoin of histone deacetylase 1 (HDAC1) is a common feature in AML and many of the HDAC1-binding altered the promoters of genes involved in hematopoiesis, transcriptional regulation and signal transduction. Furthermore, HDAC1 binding patterns were associated with patients' event free survival [45].

Esophageal and gastric cancers

Gastric cancer (GC) is a common aggressive malignancy and its prognosis remains poor in patients with more advanced stages of disease. In 2012, GC incidence in the United Sates is estimated at 21,300 new cases, while the mortality rates are estimated at 10,540 deaths [29]. Immunohistochemistry analysis of gastric adenocarcinomas showed that global levels of trimethylation of H3K9 (H3K9me3) were positively associated with tumor stage, lymphovascular invasion, and cancer recurrence. Moreover, higher level H3K9me3 correlated with a poor survival rate [46]. While a genome wide analysis of H3K27me3, which is a mark of transcriptional repression, via ChIP-ChIP, demonstrated that 128 genes displayed significant differences in H3K27me3 levels, with 119 genes exhibiting increased levels (i.e. MMP15, UNC5B, and SHH) of the mark and nine a decrease level (i.e. RB1 and AFF3) [47].

Esophageal cancer accounts for 1-3% of all cancers occurring in the United States and it is estimated that roughly 12,000 men will die of esophageal cancer in 2012 [29,48]. Clinicopathologic analysis of esophageal squamous cell carcinoma (ESCC) samples from patients who recovered from esophagectomy revealed that there was a positive correlation between tumor differentiation and H3K18ac, H4R3me2 and H3K27me3 global levels. Low expression of H3K18ac and H3K27me3 was demonstrated to correlate with better prognosis of ESCC patients, especially for those of early stages [49]. H3K18ac and H4R3me2 were also associated with recurrence-free survival (RFS) in another esophageal squamous cell carcinoma cohort. Global levels of H3K18ac was significantly associated with RFS in stage III cases and clustering analysis showed that patients with high global levels of H3K18ac and H4R3me2, had a poor RFS in stage IIB and stage III cases [48]. ESCC patients also exhibited aberrant expression of EZH2, which was significantly associated with larger size, greater depth of invasion, presence of distant metastasis, and shorter disease-free survival time [50].

Kidney, liver and pancreatic cancers

Roughly 65,000 new cases and 13,540 deaths of kidney (renal) cancer are expected to be in 2012. Kidney cancer can be classified into 3 categories including: renal cell carcinoma (RCC) (92%), renal pelvis carcinoma (RPC) (7%), and Wilms tumor (1%), a childhood cancer that usually develops before age 5 [29]. When levels of histone H3 lysine 4 (H3K4) mono, di and tri methylation were evaluated in RCC samples it was demonstrated that H3K4me3 staining was more intense in papillary RCC, whereas H3K4me1 and H3K4me2 were similar in the diverse RCC subtypes. When combined, the H3K4 methylation modifications were independent predictors of RCC progression-free survival, indicating that progression-free survival and cancer-specific survival were shorter in patients with low levels of H3K4me1-3 [51]. Samples from patients with papillary RCC also had more intense staining for H3K27 mono, di and tri methylation and progressions free survival was shorted in patients with lower levels of H3K27me1 and H3K27me3 [52]. Poor prognosis in RCC patients was also associated with lower levels of H3K9me1 [53].

An estimated 28,720 new cases of liver cancer (including intra-hepatic bile duct cancers) are expected to occur in the US during 2012, and more than 80% of these cases are hepatocellular carcinoma (HCC). HCC is one of the leading causes of cancer deaths worldwide and the long-term prognosis remains poor, despite improvements in surgical techniques and chemotherapies [29,54]. Immunohistochemistry analysis revealed that high expression of H3K4me3 was associated with poor prognosis in HCC patients [54]. H3K27me3 levels were also indicative of vascular invasion and poor patient prognosis in HCC cases. Correlation analysis demonstrated that high expression of H3K27me3 in HCCs was significantly correlated with large tumor size, multiplicity, poor differentiation, advanced clinical stage and vascular invasion. In addition, high expression of H3K27me3 in HCC patients was associated closely with shortened survival time, independent of serum α-fetoprotein levels, tumor size and multiplicity, clinical stage, vascular invasion and relapse [55].

Pancreatic cancer is a very aggressive disease with a 5-year mortality of 97-98%. The extremely poor prognosis of this malignancy is largely due to its aggressive biological behavior and late onset of symptoms for clinical diagnosis [56]. Analysis of a two large pancreatic adenocarcinoma cohorts showed that low cellular levels of H3K4me2, H3K9me2, or H3K18ac were each significant and independent predictors of poor survival in univariate and multivariate models, and combined low levels of H3K4me2 and/or H3K18ac were the most significant predictor of overall in the University of California, Los Angeles cohort. In subgroup analyses, histone levels were predictive of survival specifically for those patients with node-negative cancer or for those patients receiving chemotherapeutic agents such as adjuvant fluorouracil, but not gemcitabine [57]. Moreover, the expression of EZH2 in pancreatic ductal adenocarcinoma was also associated with patient survival and aggressiveness of disease. High levels of EZH2 were significantly correlated with decreased E-cadherin expression and more aggressive disease. Patients with low levels of EZH2 that were treated with gemcitabine also experienced significantly lower survival [58].

Conclusion

Despite recent diagnostic and technological improvements, cancer continues to retain its heavyweight status as one of the most challenging diseases to treat. It is a heterogeneous disease that often results in different clinical outcomes for patients with the same affected tissue. And as such, the disparateness of this disease makes it extremely difficult to battle. However, the ability to anticipate the clinical behavior of cancers would allow for determining the most suitable therapeutic interventions. Therefore, a considerable amount of effort is being invested in the discovery of molecular biomarkers that can categorize cancer patients with distinct clinical outcomes to expand prognostic capabilities. A better understanding of the early histological and molecular changes that occur in precursor lesions, as well as, malignant tumors, could also potentially improve clinical prognosis and allow for the development of better therapeutic agents.

As evidenced by the literature described in this review, post-translational histone modifications play an important role in cancer biology and appear to be notable predictors of disease subtypes, progression and patient survival. Interestingly, a global decrease in histone marks, such as, H3K9ac, H3K18ac, H3K4me1, H3K4me2, H3K9me2, H3K9me3, H4K5ac, H4K8ac, H4K16ac, H4K20me3, and H4R3me2 in tumor tissues, was associated with poor patient prognosis in a number of cancers. It has been suggested that this apparent decrease in histone acetylation and methylation may be a consequence of cancer cell metabolism, since their rapid proliferation and macromolecular biosynthesis may deplete the levels of acetyl and methyl donors, such as acetyl coA (AcCoA) and s-adenosyl methionine (SAM) [3].

Although, further research is needed to validate histone marks as unfailing prognostic biomarkers of cancer, the existing literature provides a compelling account of their potential. Furthermore, in addition to being prospective molecular biomarkers of disease, histone marks may also serve as potential targets of epigenetic cancer therapies [59]. Histone targeting drugs, such as, histone deacetylase (HDAC) inhibitors, suberoylanilide hydroxamic acid (SAHA, Zolinza), and romidepsin (Istodax) have already been approved for clinical use by the FDA, and many more continue to be developed and evaluated. Given their dual potential for being both biomarkers of disease and targets of anticancer therapies, the role of post-translational histone modification in cancer etiology warrants further research and exploration.