Lamarckism Revisited - Epigenetics and its Implications for Modern Health Care

1. The Roots of Epigenetics

With hindsight it is really remarkable to what significant extent Jean-Baptiste Lamarck (1744 - 1829) who also established the term ‘biology’ in 1802, has in a way contributed to modern genetics. When Charles Darwin published The Origin of Species 30 years after Lamarck’s death, he definitely considered one of his laws, the inheritance of acquired characteristics, as a supplementary mechanism to evolution in addition to natural selection. But Lamarckian evolution (a.k.a. Lamarckism) was - quite justifiably at the time - discredited by mainstream geneticists after the 1930s due to lack of a plausible rationale that was compatible with Mendelian genetics [historical context under 1].
The term epigenetic which has retained a somewhat sketchy connotation to date [2, 3, 4] was first introduced to developmental biology in 1942 by Conrad Waddington (“epigenetic landscape”) [5] who tried to reconcile Lamarck’s theory with classical genetics.

Illustration of Waddington’s epigenetic landscape

2. The Basis of Modern Epigenetics

Epigenetics takes care of the fact that chromatin (which consists half of proteins) and not just naked DNA represents the physiological template for transcription, replication, recombination and DNA repair and also constitutes the actual heritable material being passed on to the daughter cells during mitosis and meiosis. Notwithstanding the ongoing efforts to find an accurate classification for this young discipline [3] the practical definitions recently provided by Peter Jones have shaped up as the most popular so far [6]: epigenetics describes the “somatically heritable states of gene expression that are not coded in the DNA sequence itself” and epigenomics deals with “the totality of epigenetic marks in a given cell type”.
Epigenetic modifications provide a cellular memory for transcriptional control in all cell types of higher organisms by means of four mechanisms that cooperate like an integrated circuit:

1. RNA interference (miRNA, siRNA) associated silencing [reviewed elsewhere in 7, 8],
   
2. change of the chromatin structure via covalent modification of histone tails (“histone code”) [reviewed elsewhere in 9, 10, 11],
3. nucleosomal remodeling, occupancy, and turnover
4. addition of a methyl group to the 5’-carbon of cytosine in the double-stranded DNA context of palindromic 5’-CpG-3’ dinucleotides.

CpG methylation increases the affinity of DNA to histones which leads to changes in primary chromatin structure and higher nucleosome occupancy [12, 13]. By obstructing access of transcription factors to their recognition sequences, CpG methylation works as a master switch which can stably turn off cellular transcription and provides a robust mechanism for long-term silencing of unwanted transcriptional activity from tissue-specific genes and for protection against endoparasitic DNA, i.e. transposable elements [14].

DNA stretches enriched for CpG dinucleotides are called CpG islands (CGIs). Bona fide CGIs which are unmethylated in their target tissues overlap with the majority of all human gene promoters, cis-acting enhancer and silencer elements and foster the open chromatin structure that allows efficient expression of housekeeping, gatekeeper and caretaker [15] genes. The sequence-based definition of CGIs by the three Gardiner-Garden criteria [16] (as modified by Jones [17]), i.e. a length of >500 bp, a GC content ≥55%, and an observed to expected CpG ratio of ≥0.65, has recently been augmented with an empirically derived epigenetic score of CGI strength which can be used to find new active genes and distinguish them from inactive pseudogenes [18].

DNA methylation by itself is not sufficient for turning the histone code into the silencing mode: the key signal transducers are methylated-CpG-binding-proteins (MBDs) [reviewed in 19, 20] which recruit chromatin remodeling enzymes such as histone deacetylases (HDACs): the elimination of acetyl groups from ε-N-acetyllysines on histone tails facilitates the tightly packed DNA state that is characteristic for inactive heterochromatin. Different methylation densities seem to attract different types of MBDs. Although at least one MBD family member (MeCP2) has been shown to be able to bind to a single CpG pair in double stranded DNA, others bind with higher affinities but seem to require longer segments with more than one methylated CpG [21]. Nucleosomal occupancy across and directly upstream of the - often elusive - transcription start site (TSS) is dependent on CpG methylation [22, 23]. Likewise, a recent study points to a pivotal role of the three nucleosomes immediately downstream of the TSS and the CpG units in the DNA wrapped around them for long-term transcriptional silencing [24].

CpG methylation is a particularly favorable epigenetic biomarker because it can be measured as chemically stable and positive DNA signal against mostly unmethylated normal background (see Fig. 1).

Fig. 1 | Quantitative DNA analysis as a biomarker for epigenetic phenotypes

3. Epigenomes, Environment and Disease

• Links to Environment
  While it remains controversial whether ageing is the cause or effect of global changes of DNA methylation, it is obvious that the environment permanently affects the cellular biology of human beings during their whole lifetime, - by means of xenobiotics such as tobacco smoke, heavy metals, radioactivity, pollutants, infectious agents, and drugs, but also by means of diet and social environment. The striking new paradigm is that these accumulated influences can be heritable, i.e. passed on to subsequent generations by epigenetic mechanisms without changes in DNA sequence [detailed reviews in 25, 26].
  Some of the most notable examples:
• the so-called 2nd World War ‘Dutch Hunger Winter’ from 1944-1945 seemed to exert a heritable effect on the birth weight and health of the children and grand-children of the affected people [27];
• a diet rich in methyl donors (such as onions, beets, and certain vitamin mixes frequently recommended to pregnant women) administered to pregnant Agouti mice influenced coat color, body weight, and health [28] of their progeny;
• an increased stress sensitivity of adolescent rats - accompanied by sustained hippocampal transcriptome changes in their brain - could either be stimulated by low levels of maternal care (deprivation of nurturing licks by their mother rats during their infancy [29]) or by stress-induced prenatal exposure to excess glucocorticoids [30];
• differences in gene expression between monozygotic twins as they get older seem to be induced epigenetically by diverse environmental factors and lifestyles [31].
• Links to Common Disease and Mental Health
  There is a growing body of evidence that the impact of epigenetic imbalances goes far beyond the field of neo-plastic disease on which research is still mainly focused (see chapter below). Notable examples are autoimmune diseases [32], cardiovascular disease and diabetes [33], drug addiction [34], and neurodegenerative and mental diseases [35, 36]. A recent study suggests that DNA methylation plays a key role in memory formation [37] which underscores the relevance of epigenetics for neurology [a general outline of clinical epigenetics can be found in reviews 38, 39, and 40]. It is now also being increasingly recognized by geneticists that SNP-centric disease association studies have frequently fallen short of delivering novel drug and diagnostic targets due to exclusion of the epigenetic layer [41].
• Links to Ageing and Cancer
  Global CpG hypomethylation in tumor cells [42] was first described in 1983 [43], but attracted little attention at that time. The later-discovered opposite phenomenon, promoter hypermethylation of tumor suppressor and DNA repair genes, generated much more awareness, not least because it is therapeutically actionable. Andi Feinberg et al have now tried to merge the recently proposed cancer stem cell hypothesis [44] with the classic multiple hit cancer theory into an epigenetic progenitor model of cancer [45]:
  (i) latency phase: epigenetic disruption of stem and progenitor cells in their stromal niche compartment mediated by activation of tumor-progenitor genes during environmental and age-dependent exposure,
  (ii) primary tumor phase: incidence of gatekeeper mutations in tumor suppressor genes,
  (iii) tumor evolution phase: loss of homeostasis within the progenitor cell population caused by epigenetic and genetic plasticity - allowing the tumor to enter into its uncontrolled expansive state (invasion, metastasis, later on drug resistance).
  Several recent articles are supporting this model:
• precancerous stem cells have been identified and characterized in leukemic mice [46];
• an intestinal polyposis mouse model was used to establish the relationship between stem cells and tumor formation [47];
• immunity to a stem cell protein seems to protect patients with a specific precancerous condition from developing full-blown myeloma [48].

There is mounting evidence that the highly conserved repressive Polycomb Group (PcG) proteins [49, 50, 51] act as sentinels watching over cellular pluripotency or senescence. This suggests an intrinsic epigenetic mechanism for the interactivity been malignancy and mortality and also provides a source for novel targets to fight cancer [52, 53, 54, 55, 56, 57, 58].

4. Epigenetic Cancer Therapy

• DNMT Inhibitors
  CpG methylation in vertebrates is catalyzed by DNA methyltransferases (DNMT1, 3a, and 3b). Consequently, hypermethylated tumor suppressor genes are a direct result of aberrant DNMT targeting. In 1980 Peter Jones and Shirley Taylor opened the door to epigenetic cancer therapy by introducing cytidine analogues as unspecific suicide inhibitors of DNMTs which are not only able to reactivate silenced tumor suppressor genes [59] but also microRNAs that in turn can act as translational repressors of oncogenes [60]. 5-azacytidine (Vidaza®), and decitabine (Dacogen®) are the first FDA approved compounds of this class for second-line treatment of myelodysplastic syndrome and leukemias, and the less instable, oral acting chemical analogue zebularine [61, 62] is under evaluation in preclinical studies.
  
  Recently, several classic drugs and natural compounds emerged as non-nucleosidic DNMT inhibitors, such as the vasodilator hydralazine, the anesthetic procaine and the antiarrhytmic procainamide, as well as the dietary compounds genistein (and related soy isoflavones) and epigallocatechin gallate (the major polyphenol in green tea). These are now being considered directly as demethylating agents or as lead compounds for more potent and selective molecules [detailed review on DNMT inhibitors in 63].
• HDAC Inhibitors
  A lot of companies are currently focusing on the development of histone deacetylase (HDAC) inhibitors which are also effective in non-dividing tissues and can reactivate silenced tumor suppressor genes by keeping histone tails acetylated and chromatin open [63]. The first HDAC inhibitor vorinostat, a.k.a. SAHA (Zolinza™) [64] has been FDA approved in October 2006 for second-line therapy of cutaneous T-cell lymphoma. In order to avoid unwanted side effects from unspecific protein acetylation, novel structural candidates are being explored that are hoped to be more selective for some of the relevant isoforms of the Class I and II HDACs [detailed HDAC review under 65]. The drug class is currently also tested for other indications such as diabetes, inflammatory and neurodegenerative diseases as well as for antifungal therapy.
  
  Due to their synergistic mode of action, combination therapies with DMNT inhibitors are under intense investigation. Recent findings strongly suggest that the relationship between CpG methylation and histone code is not unidirectional as previously thought and HDAC inhibitors can in fact also induce DNA demethylation in non-dividing tissues such as the brain [66]. This is further evidenced by the finding that the long known antiepileptic drug valproate, more recently also used for treatment of migraine and bipolar disorder, has been shown to not only inhibit HDAC at therapeutic levels but also to reverse CpG methylation [67]. This means that DNA methylation profiles are suitable efficacy biomarkers for HDAC inhibitors where apoptosis is not the only end point of clinical interest. The valproate data on one hand suggest an epigenetic mode of action for its antidepressive effects but also alert us that central side effects may be possible in patients treated with HDAC inhibitors - in case they are lipophilic enough to cross the blood-brain barrier.
  
  Several dietary anti-carcinogenic agents such as diallyl disulfide from garlic and the isothiocyanate sulforaphane, an aglycone breakdown product of the glucosinolate glucoraphanin which is primarily found in cruciferous vegetables (broccoli and cauliflower sprouts have the highest concentrations), are now attracting the interest of researchers since it has been shown that their metabolites have a measurable HDAC inhibitory effect [68, 69] and moreover provide new clues for the molecular mechanisms of HDAC inhibition.
• HMTase Inhibitors
  Natural and synthetic and inhibitors of histone methyltransferases (HTMases), which are now emerging as alternative and potentially synergistic group of histone code modulators besides HDAC inhibitors, demonstrate the high level of interest in epigenetic cancer therapy [70, 71]. The first patents for specific compounds of this class with activity in cell lines have just been filed and it will take some time until we see the fist clinical data.
• Opportunities for Novel Combination Therapies
  Combinations of classic anti-cancer treatment schedules with epigenetic drugs are currently being added to the existing arsenal: demethylating agents are able to reverse acquired chemotherapy resistance [72, 73, 74, 75] and reactivate tumor antigens [76]. Both mechanisms open up new exciting opportunities for improved immunotherapeutic strategies [77, 78], radiotherapy [79], and classic alkylating and chemotherapeutic treatments [80, 81] when administered simultaneously with DNMT and/ or HDAC inhibitors.

5. Peeling the Onion: Mapping DNA Methylation

It is obvious that ‘the’ human epigenome doesn’t exist and virtually each of the more than 200 different human cell types possesses its individual tissue-specific and disease-specific epigenome. The ambitious goal of several national and international initiatives such as the Human Epigenome Project (HEP) or the Cancer Epigenome Project is to systematically decipher the multi-layered epigenetic patterns of human tissue and cell types and their specific signatures depending on sex, developmental stage, age, environment, disease type and state [82, 83, 84, 85, 86].

• CGI Methylation in Translational Studies
  The study of aberrant DNA methylation in cancer has quickly taken center stage in epigenetic research. CpG island hypermethylation as a key mechanism for gene inactivation has been described in almost every tumor type and it is anticipated that ongoing studies will reveal a number of valuable universal cancer methylation biomarkers which are independent of the tumor type [87]. The major clinical applications besides epigenetic cancer therapy are:
  (i) early diagnosis of occult cancers (screening),
  (ii) cancer prognosis (staging),
  (iii) therapy response (pharmacoepigenetics).
  The ultimate goal is to use circulating methylated DNA as tumor markers for non-invasive preventive medical care [88, 89, 90].
• Best Practice for Cancer Methylation Studies
  The ideal situation would be to map the methylomes of all cancer types on a single cell level for their progenitor, differentiated primary tumor, and altered disease states (metastasis). Although increasingly sophisticated microscopic 3D and 4D imaging techniques in combination with microarrays allow a first glimpse at intranuclear long distance pairing and ‘gene kissing’ events which are proceeding in localized core centers (‘transcription factories’) [91, 92, 93], the exact spatial and temporal analysis of epigenetic networks in a solitary nucleus on the single base or even single gene level is still a long way off so that we have to settle for the current two-dimensional methods.
  Before moving to precious tumor material from clinical samples, functional CGI methylation analysis should be established in well characterized cancer cell lines where the amount of material is not limited and which allow demethylation and reexpression control experiments using demethylating agents in vitro.
  In analogy to genotyping and gene expression, epigenetic research is typically approached in three phases:
  (i) establish genome-wide methylation profiles in relatively few samples. This can be done with increasingly powerful microarray-based platforms and methods, yet with inherent limitations in resolution, sensitivity and precision [reviewed in 94, 95]. But beware! Hardly anywhere in experimental biology is the GIGO aphorism (Garbage In, Garbage Out) more applicable as in microarray analysis: the key factor for a successful genome-wide differential methylation analysis is a highly sensitive and unbiased methylation recognition and enrichment method.
  Among the next generation technologies, methyl-CpG immunoprecipitation (MCIp) [96] has the potential to become a new gold standard since it combines the advantages of antibody-based enrichment techniques with the methyl-CpG-binding domain of human MBD2 which has the highest binding affinity of all known MBDs (down to very few methyl-CpG units in native dsDNA). In contrast to MeDIP [97], the method avoids a DNA denaturation step, allows separation of all present differentially methylated DNA into gradients of methylation densities and due to its low input requirements typically works without an amplification step which may introduce unwanted bias to global methylation profiles. Commercial MCIp kit formats are under development.
  (ii) verify the differentially methylated candidate promoters which result from (i) by quantitative fine-mapping of methylation levels. The method of choice should allow resolving even small differences of methylation occupancy at single CpGs in candidate gene promoters since tumor and stromal cells can almost never be fully separated in biopsy material. Due to large sample sizes, the method has to be cost effective and high throughput [reviewed in 98, 99, 100]. For this required level of sensitivity and specificity, methylated CpGs need to be made amenable to in vitro DNA amplification methods.
  
  The introduction of bisulfite conversion and bisulfite sequencing by Marianne Frommer and Susan Clark in the early 1990s [101, 102] overcame the major obstacle preventing PCR based methods from being used for methylation analysis in a restriction site independent fashion since bisulfite does virtually not react with methyl cytosine while converting unmethylated cytosine to uracil. This trick allows qualitative and quantitative read-outs from any sequence since only unmethylated cytosines undergo a C→T transition after PCR amplification while methylated ones don’t change. Although numerous optimized bisulfite conversion kits are now commercially available, this step remains the main contributor to process variability which mandates a good understanding of the necessary QC steps before setting up larger studies [103, 104].
  
  One of the most advanced, bisulfite-based locus-specific detection techniques is MALDI-TOF mass spectrometry which has several advantages over conventional bisulfite-sequencing. It can span complete CGIs in one reaction, is quantitative for single CpGs with excellent precision, and can be used at very high throughput with low running costs [105, 106] (Fig. 2). In combination with digital PCR [107, 108] this method obviates the need for cloning of individual DNAs for the assessment of allele-specific methylation pattern profiles, e.g. in loss of imprinting (LOI) studies.

Fig. 2 | Aligned Epigrams showing methylation signature differences in one of several candidate genes between leukemia samples from 10 individuals with different survival prognoses. The DNA was treated with an optimized bisulfite protocol that minimizes degradation. PCR primers not interfering with CpG sites were designed to amplify a 600bp CGI at the 5’ UTR of a leukemia-associated candidate gene. 33 CpG sites in each sample were quantified simultaneously in one single reaction with EpiTYPER®. Different colors display relative methylation changes in 10% increments (yellow = 0%, blue = 100% methylated).

(iii) detect and monitor promoter hypermethylation as diagnostic cancer biomarker with ultrasensitive, non-invasive methods from body liquids such as plasma, sputum, and exhaled breath [109, 110, 111, 112, 113, 114, 115]. Accurate prognostic markers for each neoplasia are just as important for meaningful screening modalities in a clinical or diagnostic setting as are detection methods with exquisite sensitivity and specificity [116]. The most established method is methylation-specific PCR [117], albeit its use isn’t without challenges since it requires bisulfite conversion [104], cannot be well multiplexed, and only offers relative quantitation.

Bisulfite-independent techniques such as methyl-binding PCR (MB-PCR) [118] offer the potential to enrich methylated DNA from minute amounts of input DNA and can be adapted to a range of existing methods like end-point PCR that allow high multiplexing and absolute quantification.

6. Outlook – The Sound of Silence Gets Vocal

The rapidly increasing popularity of Epigenetics has mainly been fueled by new insights into the mechanism of cancer and developmental biology but also by rapid technology advances and serendipitous discoveries of scientists from unrelated disciplines enabled by new technology.

An exploding body of scientific articles, a dedicated inter-national journal of Epigenetics [119], the election of epigenetics into the top ten emerging technologies in 2006 [120], and a growing number of devoted websites [121, 122, 123, 124, 125, 126, 127, 128] are testimony to the revolution that this discipline is bringing to science and modern society.
A recent market report by BCC Research [129] predicts that the global epigenetics/epigenomics market will grow at a 60% rate annually and reach a total value of >$4 Billion in 2012 - dominated by drug applications (65%), and followed by diagnostics and R&D.

The lesson to learn from the fact that our genetic legacy is not exclusively dependent on the hardwired information encoded in our genes, is our collective liability for coming generations – as very aptly hinted at by Science journalist Jill Neimark [130]:
“A map of many colors, with street signs so we can navigate, routes that we can choose, destinations that we can change. Maybe the gene isn't selfish. Maybe it's actually sensitive. […] and the epigenome may prove to be one of the more beautiful, delicate, subtle maps of all time”.

© Sequenom 2007 - The author can be reached at kschmidt@sequenom.com.
Download a pdf version of this article with a complete bibliography here: OverviewEpigenetics