GENE EXPRESSION AND CANCER
So far in thinking about the causes of cancer we have only discussed changes in the sequence of DNA, that is mutations. Mutations can be critical if they change the structure of a protein or the way in which genes are expressed and so impact on cell behaviour – the phenotype. However, gene expression can also be changed without alterations in the underlying sequence of DNA. What’s more, such changes can become fixed in the genome, just as if they were mutations, and be passed between generations. These are epigenetic changes, perhaps the best known of which is X-inactivation in female mammals where one of the two copies of the X chromosome is silenced in every cell. X chromosomes are inactivated by packaging them into a compressed form of DNA (heterochromatin) that suppresses gene expression. Tortoiseshell cats show this effect because each X chromosome carries a fur colouration gene (black and orange): inactivation of one chromosome produces the fur colour of the other.
In cancer, however, there are two other major routes to epigenetic effects. The first is through modification of proteins that associate with DNA and help it to fold up in the nucleus. The proteins are histones and the combination of DNA and histones makes up chromatin – so chromosomes are really chromatin. Because the structure of chromatin is critical for the transcription of genes, so is that of the histones and, like other proteins, they can be covalently modified after translation by the addition of, for example, phosphate, methyl or acetyl groups to their constituent amino acids.
The second route is by addition of methyl groups to cytosine bases next to guanine bases in DNA. Concentrations of such ‘CpG’ motifs occur in the promoter regions at the start of about half of human genes – CpG islands of between 0.5 and 4 kb in length. In normal cells, CpG islands are usually unmethylated but hypermethylation, leading to transcriptional silencing of specific genes against a background of general hypomethylation, occurs widely in human cancers. The pattern of genomic DNA methylation correlates with transcription in that transcriptionally active chromatin domains are hypomethylated and inactive regions are hypermethylated. A number of carcinogens can cause epigenetic effects and hence promote cancer.
Relatively recently a new class of regulators of mammalian cell function have emerged in the form of non-protein-coding genes. The RNAs these genes encode are short – just 18 to 24 bases in length, hence they are ‘micro RNAs’ – and their sequences are complementary to sequences in bona fide messenger RNA (that is they can form pairs with bases in the mRNA sequence in just the same way that DNA base-pairs in the double helix). When micro RNAs bind to their target mRNA the effect is to stop the message being processed by the ribosome – in other words, to inhibit protein synthesis (Fig. 1).
1. Micro RNA synthesis.
Sometimes it’s just temporarily blocked and sometimes the association leads to the mRNA being broken down – a very effective way of preventing protein translation from that message.
There are at least 800 human miRNAs and they are involved in the regulation of cellular differentiation, proliferation and apoptosis. It will come as no surprise, therefore, to learn that miRNAs are important in cancer: the pattern of expression of many of the changes in tumours by comparison with the corresponding normal tissue. Quite often the level of miRNAs is lower in tumours, which suggests that when switched on they act as a protection against tumour development. Even so, the patterns of expression vary markedly between different types of tumours – for example, leukaemias can be distinguished from solid tumours by their miRNA profiles and, for some tumours, sub-sets of miRNAs may actually be switched on as part of tumour development. In addition, miRNA networks regulate tumour metastasis and the tumour cell microenvironment.
Cancer is driven by a mutational steeplechase in which DNA damage that fails to be repaired becomes fixed in the genome. Somewhere between five and fifteen critical mutations – so-called ‘drivers’ – are thought to be necessary for most types of cancer to develop fully. Two major classes of ‘cancer gene’ contribute drivers – oncogenes and tumour suppressors. Oncogenes exert dominant effects – that is, the mutation that converts a ‘proto-oncogene’ into an oncogene confers gain-of-function. Tumour suppressor genes lose function as a result of mutation. Mutations in DNA span the entire range from a single base change through loss of large regions of DNA to complete chromosomal deletion and may also include shuffling of segments giving rise to new genes (chromosomal translocation). Conversely, regions of DNA or entire chromosomes may be duplicated giving rise to ‘gene amplification’. Some mutations have no effect on function. However, even a single base change, if it affects a critical amino acid can produce a ‘driver’, as it does in activating RAS in about 20% of human cancers. Gene or chromosome loss that involves a tumour suppressor removes one of the brakes on cell proliferation. Two major tumour suppressors that are frequently inactivated in human cancers are P53 and the retinoblastoma gene RB1. The juxtaposition of normally distant regions of the genome can generate novel proteins if the gene fusion event maintains the coding sequence ‘in-frame’. Chimeric proteins thus produced are a distinctive feature of leukaemias but they also occur in solid tumours. More subtle effects on cancer development may arise from polymorphisms that may, for example, slightly modulate the efficiency of gene transcription. Micro RNAs regulate protein synthesis from mRNAs and can exert either oncogenic or a tumour suppressive effects.