The role of epigenetics in cancer

By Dr. Andy Bannister *

Key epigenetic processes & links to cancer by Dr. Andy Bannister (Cambridge University)

Contents

  1. DNA methylation
  2. Histone covalent modification
  3. Genetic and epigenetic interplay towards cancer
  4. DNA methylation and cancer
  5. Histone modifications and cancer
  6. An epigenetic role for RNA
  7. References      
  8. Useful links

Epigenetics refers to heritable changes in gene expression that occur without alteration in DNA sequence. There are two primary and interconnected epigenetic mechanisms - DNA methylation and covalent modification of histones.  In addition, it is also becoming apparent that RNA is intimately involved in the formation of a repressive chromatin state. Here I review some of the many potential links between epigenetic processes and cancer.

DNA methylation

DNA methylation in humans occurs almost exclusively at CpG dinucleotides and most CpG sequences in the genome are methylated1. DNA methylation is catalysed by a family of DNA methyltransferase enzymes (DNMTs). The modification is generally repressive to transcription and helps to maintain silencing of transposable elements, therefore enhancing genome stability. However, there are 'CpG islands' associated with gene promoters that escape DNA methylation1.

Histone covalent modification

A plethora of covalent post-translational modifications of the histone tails have been documented (Figure 1).  The best characterized of these are acetylation, methylation and phosphorylation2,3.  For each modification, enzymes exist which either lay down the appropriate mark or remove it (except for arginine methylation where the only known 'demethylase' demethyliminates mono-methyl arginine to form citrulline) (Figure 1).  Active genes tend to be enriched with particular modifications (e.g. H3K4me3 and H3K9ac) whereas inactive genes have different specific enrichments (e.g. H3K9me2 and H3K9me3 and H3K27me2 and H3K27me3).  However, there are no absolute rules and many active and inactive genes have overlapping patterns of histone modifications. Nevertheless, it is clear that aberrant regulation of histone modification can affect gene activity and therefore oncogenic potential exists2,3.

Figure 1 - Epigenetic modifications

Epigenetics01
Figure 1.  The dynamic nature of epigenetic modifications.  The modifications that occur on histone N-terminal tails and on DNA are shown together with the enzymes that lay down and remove the marks.  Deregulation of any of these enzymes has the potential to be oncogenic.

Genetic and epigenetic interplay towards cancer

Much effort has been invested in identifying genetic mutations in cancer.  In inherited cancer syndromes this approach has proved successful.  Furthermore, mutations early in the genesis of common cancers have also been identified and these are likely to be associated with tumour initiation.  In contrast, few specific genetic mutations have been linked to tumour progression, leading Feinberg to suggest that epigenetic changes may be involved4.  Epigenetic changes occur without a change in the DNA sequence and they can be induced by various factors (see below).  Thus it is possible, for example, that a DNA mutation leads to cellular transformation, but induced changes in the epigenome of the transformed cell enhances the probability that it will be capable of metastasising4 (Figure 2).  In this scenario, a genetic mutation initiates the cancer but epigenetic change promotes its progression.

Epigenetic processes may also be involved in cancer initiation.  It is possible that epigenetic change may lead directly to cancer initiation (Figure 2).  Alternatively, changes already induced within the epigenome may 'prime' cells in such a way as to promote cellular transformation upon a subsequent DNA mutagenic event4 (Figure 2).  In this case the epigenetic component of the cancer initiation is intricately entwined with the genetic component.  The involvement of epigenetic change in cancer initiation is of course not mutually exclusive to it having also a role in cancer progression as discussed above (Figure 2).

Figure 2 - Epigenetic alterations in the genesis of cancer

Epigenetics Fig 1
Figure 2.  How genetic and epigenetic alterations may cooperate in the genesis of cancer. Potential pathways are shown indicating how genetic change may precede epigenetic change, and vice versa, as the cause of cancer - see text for details.

A genetic alteration in the gene encoding an 'epigenetic enzyme' (e.g. a histone acetyltransferase - see below) may lead to changes within the epigenome.  If, for example, these changes cause the activation of an oncogene then cancer may arise.  In addition, mutations in genes that code for proteins that recognize and bind to epigenetic marks (e.g. methyl binding domain proteins and bromo/chromo domain proteins which bind to methylated DNA and acetylated/methylated histones respectively) could be as important in cancer as mutations in the enzymes themselves.  Although these are genetic events that lead to cancer, an alteration in the epigenome most likely also plays a part.  However, it should be noted that many of the histone-modifying enzymes also modify non-histone proteins, thus making a direct link between enzyme deregulation, changes in the epigenome and cancer extremely difficult.

DNA methylation and cancer

DNA methylation was the first epigenetic alteration to be observed in cancer cells5. Hypermethylation of CpG islands at tumour suppressor genes switches off these genes, whereas global hypomethylation leads to genome instability and inappropriate activation of oncogenes and transposable elements1. It appears that genomic DNA methylation levels, which are maintained by DNMT enzymes, are delicately balanced within cells; over-expression of DNMTs is linked to cancer in humans, and their deletion from animals is lethal5,6.  Furthermore, methyl cytosine is capable of spontaneously mutating in vivo by deamination to give thymine.  Indeed, 37% of somatic p53 gene mutations (and 58% of germ-line mutations) occur at methyl CpGs, and these mutations are strongly implicated in the cause of cancer7.

It is important to note that there is a direct link between DNA methylation and histone modification.  A number of proteins involved in DNA methylation (e.g. DNMTs and MBDs) directly interact with histone modifying enzymes such as histone methyltransferases (HMTs) and histone deacetylases (HDACs)8.  In fact, it is believed that DNA methylation and histone methylation are tied together in a reinforcing loop where one modification depends on the other.  Upsetting this relationship will almost certainly have severe consequences on the epigenome and chromatin organization.  Thus most, if not all, factors that affect DNA methylation levels will also affect histone modifications.  For instance, it seems that H3K9 methylation and DNA methylation are linked8.  Moreover, in cancer cells, disruption of DNA methylation is linked to loss of H4K20 methylation9.

A number of factors can influence the DNA methylation levels of a cell without requiring a change in genomic DNA sequence. 

  • Aging: With aging in certain tissues there is a general tendency for the genome to become hypomethylated whereas certain CpG islands become hypermethylated, a situation reminiscent of that found in many cancer cells10.  Whether this age-dependent change in DNA methylation is linked to the increased cancer incidence in later life remains to be determined. 
  • Diet:  Nutrition supplies the methyl groups for DNA (and histone) methylation via the folate and methionine pathways.  Importantly, mammals cannot synthesise folate or methionine and so a diet low in these compounds leads to alterations in DNA methylation. These changes have been associated with cancer6.
  • Environment:  Agents such as arsenic and cadmium can have profound effects on DNA methylation.  Arsenic causes hypomethylation of the ras gene whereas cadmium induces global hypomethylation by inactivating DNMT1ref 11,12.

DNA methylation is also the principal epigenetic factor governing allelic imprinting.  Imprinting is the process by which only one allele of certain genes is expressed depending on the parental origin.  In other words, a gene 'remembers' whether it was inherited from the father or the mother and is expressed accordingly.  DNA methylation is part of this 'memory' process and loss of imprinting (LOI) has been tightly linked to cancer susceptibility.  For example, the insulin-like growth factor -2 (IGF2) gene is an imprinted gene that is expressed only from paternally-inherited alleles.  However, aberrant methylation of the IGF2 imprinting control region (ICR) on the maternal allele leads to additional expression of the maternal copy.  Thus, the levels of IGF2 rise approximately two-fold and this elevated level is associated with cancer susceptibility4,5

Histone modifications and cancer

The histone N-terminal tails are crucial in helping to maintain chromatin stability and they are subject to numerous modifications.  Most modifications have some role to play in transcriptional regulation and so each has the potential to be oncogenic if deregulated deposition leads, for example, to loss of expression of a tumour suppressor gene2,3

  • Acetylation:  Histone acetylation tends to open up chromatin structure.  Accordingly, histone acetyltransferase (HATs) tend to be transcriptional activators whereas histone deacetylases (HDACs) tend to be repressors.  Many HAT genes are altered in some way in a variety of cancers2,3.  For instance, the p300 HAT gene is mutated in a number of gastrointestinal tumours.  On the other hand, alteration of HDAC genes in cancer seems to be far less common.  However, despite this low incidence of genetic mutation in cancer, HDAC inhibitors are performing well in the clinic as anti-cancer drugs.
  • Methylation:  All lysine methyltransferases that target histone N-terminal tails contain a so-called SET domain.  This domain possesses lysine methyltransferase activity and numerous SET domain-containing proteins are implicated in cancer13,14.  One example is the Suv39 family of enzymes that catalyse methylation of H3K9.  Transgenic mice devoid of these enzymes are very susceptible to cancer, especially B cell lymphomas.  Histone demethylases have only very recently been identified and as yet no linkage to cancer has been observed.  However, such a linkage seems probable.
  • PhosphorylationH3S10 and H3S28 are phosphorylated at mitosis - a crucial part of the cell cycle; misregulation here is often associated with cancers.  Indeed, the Aurora kinases that perform this H3 phosphorylation are implicated in cancer2,3

An early event following DNA damage is the phosphorylation of H2AX, a process that is required for efficient DNA repair.  If repair is not performed correctly the cell is left with damaged DNA, with predictable consequences.

An epigenetic role for RNA

RNA can also be regarded as an epigenetic component involved in chromatin regulation.  For example, in female mammals the noncoding Xist RNA helps to silence the inactive X-chromosome by coating the chromosome and promoting the formation of heterochromatin.  The RNA interference (RNAi) pathway is also linked to chromatin structure; disruption of components of the RNAi machinery affects the formation of heterochromatin.  Whether these processes can be directly linked to cancer remains to be determined.  However, in the case of micro RNA (miRNA) the link is strong.  Micro RNAs affect the expression of genes linked to the cell cycle (eg down-regulation of E2F1) and expression of miRNAs is altered in cancer cells15. Furthermore, miRNA profiling has proved to be a very useful aid to classifying different cancer types15.

Unlike genetic alterations, epigenetic changes are potentially reversible. The large-scale development of small molecule inhibitors of DNA and histone-modifying enzymes is in now in full swing.  In the clinic, the success of HDAC inhibitors and DNA demethylating agents like aza cytidine as anti-cancer drugs demonstrates 'proof of principle' of this approach and provides great hope for the development of a more comprehensive portfolio of 'epigenetic drugs' in the future.

References

  1. Egger G, Liang G, Aparicio A, Jones PA. Epigenetics in human disease and prospects for epigenetic therapy.  Nature (2004) 429:457-463
  2. Zhang K, Dent SY. Histone modifying enzymes and cancer: going beyond histones.  J. Cell. Biochem. (2005) 96:1137-1144.
  3. Santos-Rosa H,  Caldas C. Chromatin modifier enzymes, the histone code and cancer.  Eur. J. Cancer (2005) 41:2381-2402.
  4. Feinberg AP. The epigenetics of cancer etiology.  Seminars Cancer Biol. (2004) 14:427-432.
  5. Feinberg AP,  Tycko B. The history of cancer epigenetics.  Nat. Rev. Cancer (2004) 4:143-153.
  6. Rodenhiser D, Mann M.  Epigenetics and human disease: translating basic biology into clinical applications.  C.M.A.J. (2006) 174:341-348.
  7. Rideout WM 3rd, Coetzee GA,  Olumi AF, Jones PA. 5-Methylcytosine as an endogenous mutagen in the human LDL receptor and p53 genes.  Science (1990) 249:1288-1290.
  8. Fuks F. DNA methylation and histone modifications: teaming up to silence genes. Curr. Opin. Genet. Dev. (2005) 15:490-495.
  9. Fraga MF et al. Loss of acetylation at Lys16 and trimethylation at Lys20 of histone H4 is a common hallmark of human cancer.  Nat. Genet. (2005) 37:391-340.
  10. Richardson B. Impact of aging on DNA methylation.  Ageing Res. Rev. (2003) 2:245-261.
  11. Okoji RS, Yu RC, Maronpot RR, Froines JR. Sodium arsenite administration via drinking water increases genome-wide and Ha-ras DNA hypomethylation in methyl-deficient C57BL/6J mice.   Carcinogenesis (2002) 23:777-785. 
  12. Takiguchi M, Achanzar WE, Qu W, Li G, Waalkes MP. Effects of cadmium on DNA-(Cytosine-5) methyltransferase activity and DNA methylation status during cadmium-induced cellular transformation.  Exp. Cell Res. (2003) 286:355-365.
  13. Lund AH, van Lohuizen M. Epigenetics and cancer.  Genes Dev. (2004) 18:2315-2335.
  14. Schneider R, Bannister AJ, Kouzarides T. Unsafe SETs: histone lysine methyltransferases and cancer. Trends Biochem. Sci. (2002) 27:396-402.
  15. Mattick JS, Makunin IV. Non-coding RNA. Human Molec. Genet. (2006) 15:R17- R29.

Useful Links


Dr. Andy Bannister
Gurdon Institute, University of Cambridge