Enabling Technologies
3D nuclear DNA methylation patterns in cell-based assays, Cell-by-cell characterisation of cell populations
Utilising 3D nuclear DNA methylation patterns in cell-based assays for epigenetic drug screening
By Professor Jian Tajbakhsh, Professor Arkadiusz Gertych and Professor Daniel L Farkas
Spring 2010

Since epigenetic changes are thought to underlie a wide range of complex diseases, the scope of epigenetic drug therapy is likely to expand, as epigenetic phenomena, in contrast to pure genetic mechanisms, are reversible.

The boost in epigenetic drug development calls for high-throughput cell-based screening assays to be implemented into the early phases of drug discovery for assessing the agents’ genotoxic side-effects on chromatin architecture and the correlated risk of genome instability in targeted cells and tissues. The authors of the article introduce a novel cytometric approach termed 3D Quantitative DNA Methylation Imaging, which addresses this matter.

The technology applies a machine-learning algorithm for extraction of fluorescent signals from threedimensional images of chromatin texture to visualise and measure druginduced changes in global DNA methylation and related chromatin reorganisation in nuclei of thousands of cells in parallel. The method is amenable to scale and can be flexibly implemented for drug screening in research and in the pharmaceutical industry.

Pharmaceutical drugs act on cellular processes or pathways to induce physiological changes. In addition drugs may also exert epigenetic effects, which affect gene expression and related chromatin architecture in cells. These latter effects can be targeted but also occur as unwanted adverse occurences (side-effects) that might put cells, organs and eventually the whole organism at risk (1).

Our rapidly accumulating knowledge of such possible changes in the current therapeutic practice calls for strategies and systems biology approaches in drug screening and drug discovery that can detect unwanted epigenetic effects, in order to improve the agents’ clinical utility. These new approaches, which have been termed pharmacoepigenomics or toxicoepigenomics (in analogy to pharmaco-/toxicogenomics), if introduced into the drug development programme, might also significantly reduce attrition in the early phases of drug development.

Cancer epigenetics and drug treatment

DNA and histones are the basic components of a chromosome, in which the DNA helix is wrapped around core histones to form the beads-on-a-string structure that is then folded into higher-order chromatin. Epigenetic changes, such as DNA methylation and histone modifications, are key regulators of gene expression and chromatin architecture in a cell- and tissue-specific manner.

Aberrant global methylation patterns are associated with complex diseases including several cancer types. Methylation pattern imbalances in cancer cells include genome-wide hypomethylation and localised aberrant hypermethylation of CpG dinucleotides (CpG islands) in promoter regions of tumour suppressor genes (2). Epigenetic control of gene expression is a rapidly developing field of substantial potential, and oncology is likely to be the therapeutic application in which the fastest progress is made.

The reversible nature of epigenetic imbalances in various types of cancers constitutes an attractive therapeutic target – the disruption of epigenetic misregulation via enzyme inhibition has considerable pharmacological impact that fuels the development of epigenetic therapies by involving the use of DNA methyltransferase (DMNT) inhibitors and histone deacetylase (HDAC) inhibitors (3).

DNA methylation is considered the most robust among the different types of chromatin modifications, and researchers are now comfortable with the theory that DNA methylation can act as a causal factor in transcriptional inactivation.

In contrast, histone modifications such as acetylation and methylation are considered to be more ‘fuzzy’. However, they are believed to be more finely tuned with subtler but just as far-reaching impacts on gene expression (4). Because DNA methylation plays an important role in controlling cellular gene expression programmes, the DNA methylation machinery is an attractive pharmacological target.

Drugs that cause demethylation and the reactivation of ectopically methylated genes might be superior to many other therapeutic approaches that are targeted at specific functions of individual proteins. DNMT1 is a multifunctional protein that has regulatory domains that potentially regulate activities related to cell growth in addition to DNA methylation. The latter activity is responsible for maintenance of the DNA methylation patterns during genome replication prior to cell division.

A number of DNA methylation inhibitors of different categories have been designed, with 5-azacytidine (5-AZA) and its analogue 5-aza-2-deoxycytidine (decitabine) being approved for treatment of patients with myelodysplatic syndrome (MDS) (5). These enzymes do not remove methyl groups from methylated chromatin, but rather prevent methylation of daughter DNA during DNA replication.

Recently a new type of cytosine analogue, 1--Dribofuranosyl- 2(1H)-pyrimidone, also known as zebularine, has shown promising results for oral administration. This drug seems to be less toxic to cultured cells than the other two azanucleosides (6), and has a significantly stronger anti-proliferative effect on cancer cells than on normal cells (7).

Nevertheless it is important to recognise that hypermethylation of single gene promoters occurs against a strong background of general DNA hypomethylation including a decrease in the methylation load of heterochromatic regions of the genome. The loss of methyl groups is achieved

mainly by hypomethylation of heterochromatinresiding repetitive DNA sequences, including transposable elements such as long interspersed nuclear elements (LINE), especially LINE-1 as part of the facultative heterochromatin, and satellite 2 (Sat2) DNA as part of the constitutive heterochromatin (8). The extent of genome-wide hypomethylation converges closely with the degree of malignancy, in a tumour type-dependent manner.

In normal cells, the activity and interaction of these classes of DNA with neighbouring chromatin regions (in the nuclear space) are strongly suppressed by methylation and compaction through histone modifications. Therefore, there are multiple risks associated with an increase in demethylation of repetitive elements, which can lead to their decondensation and related consequences – Sat2 DNA hypomethylation may favour pericentromeric instability, and LINE-1 elements may become transcriptionally active and relocate into other parts of the genome causing enhanced genome instability through mutational effects such as silencing of tumour-suppressor genes or the activation of oncogenes.

This is in fact a critical issue, as malignant cells already contain 20-60% less genomic methylcytosine than their normal counterpart. Inhibitors that disrupt the cell cycle-promoting activities of DNMT1 and induce a more selective demethylation of specific genes for their reactivation without necessarily affecting heterochromatin methylation might significantly improve their clinical utility. Such second-generation epigenetic drugs, which cause an orchestrated reversal of multiple functions, are mostly in the discovery phase.

In addition to inhibitors of DNA methylation numerous histone deacetylase (HDAC) inhibitors are undergoing preclinical testing as single agents as well as in combination with DNMT inhibitors for cancer therapy. Lines of evidences have shown that combinations of these two classes of drugs exert synergistic effects on gene expression and tumour growth (9).

HDAC inhibitors are immediately effective in cells even before the following cell cycle. Nevertheless, their effects on cancer cells in vitro may ultimately also lead to DNA demethylation as a consequence of inhibition of histone deacetylation (10,11). DNMT inhibitors and HDAC inhibitors have raised high hopes for epigenetic therapies but are far from perfect. Their clinical successes have mainly been against haematological cancers, and they show little consistent efficacy against solid tumours.

Cell-based assays and cellular imaging of chromatin texture

Despite the extraordinary degree of complexity in chromatin modifications and signalling, which makes the identification of suitable compounds challenging, there is substantial enthusiasm for the development and implementation of epigenetic therapies in several human disease areas, especially oncology. Chromatin phenotype is known to be significantly disrupted in cancer. This has been demonstrated in many morphologic studies on cancer by the application of digital texture analysis for the quantitative analysis of chromatin phenotype in neoplasia – supporting the role of chromatin phenotype as a biomarker for diagnosis and prognosis.

Given the prevalence and load of DNA methylation imbalances, especially hypomethylation of repetitive elements, cellular imaging of global nuclear DNA patterns may provide a powerful tool to characterise mammalian cells during differentiation and in their status of health versus disease, as the underlying molecular processes involve large-scale chromatin reorganisation, visible by light microscopy (12).

Cellular imaging can be defined as the use of a system/technology capable of visualising a cell population, single cell or subcellular structures, applied in a combination with image-analysis tools. With the availability of today’s more advanced imaging approaches (including confocal laser scanning microscopy, two-photon excitation microscopy, high content cell imaging, and automated digital tissue scanning), high resolution optical imaging has evolved into an essential tool for moving new chemical entities through the pharmaceutical discovery pipeline utilising cell-based assays.

Imaging advantages for drug discovery are realised through the ability of high-resolution microscopic imaging to measure the spatial and temporal distribution of molecules and cellular components, which is vital to understand the activity of drug targets at the cellular level. Structureactivity relationships are commonly used by medicinal chemists to improve small-molecule properties. Cellular imaging, when combined with systems biology could promote a more biology-driven environment for compound progression, therefore compelling us to think in terms of structure-phenotype relationships.

Today’s cellular imaging systems produce multispectral two- and three-dimensional (2D and 3D) data in quantities that often require machine vision support to assess and quantify the degree of individual cell similarity within an entire cell population based on cellular features. Topological analyses typically necessitate the segmentation of cellular regions of interest (ROI), including the entire cell and/or subcellular compartments such as the nuclei. This process involves the delineation of the ROI, recognition of residing patterns, and statistical quantification of these patterns with dedicated machine-learning algorithms.

3D quantitative DNA methylation imaging: a novel cytometric approach

Since heterochromatin decondensation bears a serious risk in epigenetic therapy, it is important to consider the testing of new compounds and drugs using cell-based assays in the pre-clinical phases of drug discovery. Logically, these dedicated assays need to aim at the quantitative characterisation of global DNA methylation changes in a highthroughput statistical fashion to verify unwanted effects such as increased hypomethylation of repetitive elements and its causal risk of heterochromatin decondensation, which can cause severe adverse reactions of cells.

Therefore, we have recently developed and introduced 3D Quantitative DNA Methylation Imaging (3DqDMI), a novel image-cytometrical approach that can measure two important parameters of DNA methylation changes: (i) the methylcytosine (MeC) load, and (ii) the spatial nuclear codistribution of MeC and global DNA (gDNA). Nuclear MeC sites can be visualised with an immunoassay using a primary mouse monoclonal antibody that specifically recognises 5-methylcytosine (EMD Biosciences, Abcam) and a secondary polyclonal antibody that is fluorophore conjugated (Invitrogen).

We have demonstrated that the codistribution patterns of these two classes of DNA can be utilised as signatures for the characterisation of stem cells in differentiation and cancer cells upon treatment with demethylating agents (13) (Figure 1).

Figure 1 Effect of 5-AZA on the chromatin of AtT20 pituitary tumour cells in culture

Using our novel algorithm we succeeded in tracking changes of higher-order DNA organisation due to druginduced demethylation on a genomic scale in mammalian tumour cells (14).

Approximately 55% of the human genome is comprised of heterochromatin. Epigenetic drugs with demethylating effects have been shown to alter genome organisation within mammalian cell nuclei15. Therefore, we used the genome and its 3D organisation as an indicator to evaluate the effect of demethylating drugs on cancer cells in situ.

Our approach provides a statistical measurement on the two classes of DNAs (MeC and DAPIpositive gDNA) as nuclear targets. The algorithm compares the relative distribution of signals derived from these two targets (from two colourchannels), projects them on to scatter plots and then measures the degree of similarities between the plotted signal distributions of cells within a population (Figure 2).

Figure 2 Cell-by-cell characterisation of cell populations by 3D-qDMI for pharmacoepigenomics

This method offers a way to evaluate cellular response to external factors such as drugs and changes in culture conditions via dissimilarity assessment of relevant cellular structures. The algorithm developed combines the three major tasks: (1) automated segmentation of nuclei in a cell population, (2) subsequent nuclear pattern extraction, and (3) distance-based statistical measurement of cell similarity using Kullback-Leibler (K-L) divergence (Figure 3).

Figure 3 Structure-based homogeneity assessment of treated cell populations for assessment of drug efficacy and genotoxicity

This method considers the strength of statistical evaluation of intranuclear MeC/DAPI patterns, especially valuable when cell population homogeneity is difficult to be assessed due to lack of standardised reference and sample size.

Demethylating agents cause structural reorganisation of the genome in cell nuclei, as they not only alter the DNA methylation load but also influence its spatial distribution (12-14) (Figure 1). In particular, heterochromatin decondensation, as a secondary effect of global demethylation, results in the relocation of heterochromatic sites within the nucleus (which is associated with genome destabilisation).

As a consequence of these conformational and organisational changes of the DAPIpositive nuclear sites, the same DAPI signal intensity is spread out over a higher number of voxels. Thus, both MeC and DAPI have dynamic patterns in the cell nucleus that become more discernable in a joint 2D plots than in a 1D MeC plot, or even when the two signals are separately displayed in one dimension (14).

Kullback-Leibler’s divergence is a valuable measure for quantitating dissimilarities within a cell population and this measure can be applied to any multi-colour cellular assay that utilises topological information of intracellular structures to assess cellular behaviour. The image analysis system can flexibly interrogate the cells with any number of targets and large amounts of 3D image stacks for high-content analysis (Figure 4).

Figure 4 The 3D-qDMI system currently offers three features for the in situ assessment of drug-induced changes in global DNA methylation in cells

3D-qDMI connects fluorescence techniques including immunocytochemistry and fluorescence in situ hybridisation (FISH) to computational techniques for image analysis and data interpretation. Our experimental results underline the robustness of the method, but also its flexibility in dealing with a high dynamic range in sample size suitable for low to high-throughput cell-based screening.

The original software version, written in MATLAB (The MathWorks, Natick, MA), is compatible with Windows 7 operational system (Microsoft Inc, Redmond, WA) that we run on powerful workstations such as the 8-core Mac Pro (Apple Inc, Cupertino, CA). The system is amenable to scale and can be implemented for biomedical research as well as the high-volume industrial routine. EpiLumina, Inc (Encino, CA) is working on proprietary, optimised hardware/ software solutions for further increasing the system’s performance.


Our study of spatial DNA methylation patterns utilising a new method we developed (3D-qDMI) indicates that these signatures can serve as a phenotypic indicator for demethylation effects (13,14), as suggested by previous investigations (15,16).

These report that induced changes in global DNA methylation can influence the homeostasis of heterochromatin methylation, specifically leading to hypomethylation of repetitive elements and potentially causing an adverse reorganisation of the genome with side-effects, such as transcriptional activation of oncogenes, activation of latent retrotransposons, chromosomal instability (17), and telomere elongation of chromosomes (18).

We conclude that epigenome cell-based screening of the epigenome, especially the methylome of cells with holistic methods such as 3DqDMI, are valuable tools in the identification of potential impacts of second generation epigenetic drugs in development (that act more DNA sequence-specifically), as well as of drugs already in clinical practice. Furthermore, this type of toxicoepigenomic assessment could be practically expanded to any drug beyond anti-cancer drugs, for which we have no current knowledge of epigenetic side-effects. DDW


Professor Jian Tajbakhsh is an Assistant Professor in the Department of Surgery and Head of the Translational Cytomics Program at the Cedars- Sinai Medical Center (CSMC), Los Angeles (CA). Prior to joining CSMC he was a Principal Investigator within the biotech industry developing miniaturised and automated molecular diagnostics assays and systems, funded by multiple awards through the NIH and the NSF. He has also been a member of the 3D-Genome Study Group in Heidelberg (Germany) pioneering the utilisation of gene topology and chromatin texture in the in situ characterisation of cells.

Professor Arkadiusz Gertych is an Assistant Professor in the Department of Surgery at CSMC, an expert in bioimage informatics tool development for computer-aided diagnosis (CAD) and high-content analysis. He has also been an Assistant Professor in the Department of Biomedical Electronics at the Silesian University of Technology (Poland) and a key contributor in the development of CAD-tools for bone age assessment at the Image Processing and Informatics Lab at the University of Southern California (USC).

Professor Daniel L. Farkas is Professor of Surgery and Biomedical Sciences at CSMC, Director of the Minimally Invasive Surgical Technologies Institute (MISTI) since 2002, Research Professor in Biomedical Engineering at USC, and Adjunct Professor of Robotics at Carnegie Mellon University (Pittsburg). He has authored more than 100 peer-reviewed publications and chaired 24 international scientific meetings. He is a member of the Executive Committee of the Biomedical Optics Society, was on the editorial boards of Molecular Imaging, Cytometry, Journal of Biomedical Optics, Current Analytical Chemistry and Journal of Microscopy and has served on more than a dozen NSF panels and NIH Study Sections, and on scientific advisory boards of national research centers as well as companies.


1 Csoka, AB, Szyf, M (2009). Epigenetic side-effects of common pharmaceuticals: a potential new field in medicine and pharmacology. Med Hypotheses 73, 770-780.

2 Jones, PA, Baylin, SB (2007). The epigenomics of cancer. Cell 128, 683-692.

3 Yoo, CB, Jones, PA (2006). Epigenetic therapy of cancer: past, present and future. Nat Rev Drug Discov 5, 37-50.

4 Best, JD, Carey, N (2010). Epigenetic opportunities and challenges in cancer. Drug Discov Today 15, 65-70.

5 Luebbert, M (2000). DNA methylation inhibitors in the treatment of leukemias, myelodysplastic syndromes and hemoglobinopathies: clinical results and possible mechanismas in action. Curr Top Microbiol Immunol 249, 135-164.

6 Marquez, VE et al (2005). Zebularine: a unique molecule for an epigenetically based strategy in cancer chemotherapy. Ann N Y Acad Sci 1058, 246-254.

7 Balch, C et al (2005). Antimitogenic and chemosensitizing effects of the methylation inhibitor zebularine in ovarian cancer. Mol Cancer Ther 4, 1505- 1514.

8 Ehrlich, M (2005). DNA Methylation and Cancerassociated Genetic Instability. Chapter 3.6 in: Genome Instability in Cancer Development, Nathan Back, Irun R. Cohen, David Kritchevsky, Abel Lajtha, Rodolfo Paoletti and Erich A. Nigg, Editors, ISSN 0065-2598.

9 Hurtubise, A, Momparler, RL (2006). Effect of histone deacetylase inhibitor LAQ824 on antineoplastic action of 5- Aza-2'-deoxycytidine (decitabine) on human breast carcinoma cells. Cancer Chemother Pharmacol 58, 618-625.

10 Detich, N et al (2003). Valproate induces replicationindependent active DNA demethylation. J Biol Chem. 278, 27586-27592.

11 Ou, JN et al (2007). Histone deacetylase inhibitor Trichostatin A induces global and gene-specific DNA demethylation in human cancer cell lines. Biochem Pharmacol. 73, 1297-1307.

12 Espada, J, Esteller, M (2007). Epigenetic control of nuclear architecture. Cell Mol Life Sci 64, 449-457.

13 Tajbakhsh, J et al (2008). Characterization of tumor cells and stem cells by differential nuclear methylation imaging. In: Farkas DL, Nicolau DV, Leif RC. Editors. Imaging, Manipulation, and Analysis of Biomolecules, Cells, and Tissues San Jose: Proceedings of the SPIE Vol.; 6859F1-10.

14 Gertych, A et al (2009). Automated quantification of DNA demethylation effects in cells via 3D mapping of nuclear signatures and population homogeneity assessment. Cytometry A 75, 569-583.

15 Haaf, T (1995). The effects of 5-azacytidine and 5-azadeoxycytidine on chromosome structure and function: implications for methylation-associated cellular processes. Pharmacol Ther 65,19-46.

16 de Capoa, A et al (1996). Cytological evidence for 5-azacytidine-induced demethylation of the heterochromatic regions of human chromosomes. Chromosome Res 4, 271-6.

17 Clark, SJ, Melki, J (2002). DNA methylation and gene silencing in cancer: which is the guilty party? Oncogene 21, 5380-5387.

18 Vera, E et al (2008). Epigenetic regulation of telomeres in human cancer. Oncogene 27, 6817-6833.