High-Content Analysis with Cellular and Tissue Systems Biology: A Bridge between Cancer Cell Biology and Tissue-Based Diagnostics


Acknowledgments

This work was supported in part from research grants from the State of Pennsylvania Tobacco Funds (DLT), NIH R01 GM075205IH (RFM), NIH-NIGMS R01 GM086238 (TL, Ivet Bahar, PI), NIH-NCRR UL1RR024153-05 (TL and JRF, Steve Reiss, PI), and NIH-NCI 1R21CA164433-01A1 (CC, Liu, PI), and Pennsylvania Department of Health Cure Program Grant RFA#10-07-03 (RC-T). The authors thank Andy Stern for discussions on cancer biology and drug discovery, Adrian Lee and Steffi Oesterreich for discussions on cancer biomarkers, Ivet Bahar for discussions on computational biology, Jennifer Grandis for insights in head and neck cancer and Cal33 cells used in creating Figure 25-3 , and Kumiko Isse and Anthony J. Demetris for supplying the images used for the analysis shown in Figure 25-7 .

High-Content Analysis (HCA)

Background on Fluorescence Imaging in Cancer Biology, Drug Discovery, and Diagnostics

Cancer displays significant genetic and nongenetic heterogeneity. Tumors are integrated tissue systems of interacting malignant cells, stem cells, and stromal components, including immune cells, fibroblasts, endothelial cells, and nerve cells ( Figure 25-1 ). Immune cells and other stromal components play cooperative roles in tumor development and metastasis and influence responses to therapies. The stromal phenotype and functions are strongly associated with disease progression and clinical outcome in cancer. Leukocytes are attracted into tumors by chemokines and can both protect the tumor from antitumor immunity and promote tumor progression via stimulating angiogenesis and tumor cell migration. Tumors can render infiltrating immune cells anergic/nonresponsive or drive such cells into apoptosis. Tumor cell heterogeneity, the complexity of the tumor system, and the vital interactions of tumor cells with multiple components of the stroma highlight the need for a “tissue systems biology” approach to cancer diagnostics, which combines multiplexed biomarker measurements in the context of the tissue architecture and tumor cell function ( Table 25-1 )

Table 25-1 includes citations for References .

with informatics tools to classify individual patients according to disease subtype, recurrence, and responses to therapies.

Figure 25-1, Solid tumors are systems within the human system

Table 25-1
Key Tumor System Processes and Biomarkers
Adapted from Critchley-Thorne RJ, Miller SM, Taylor DL, et al. Applications of cellular systems biology in breast cancer patient stratification and diagnostics. Comb Chem High Throughput Screen . 2009;12:860-869.
System Process Example Biomarkers References
Proliferation Ki-67, Aurora A kinase
Apoptosis p53, Apo-1/Fas, FasL, TRAIL receptors, caspases, pAKT, Survivin, MCL-1, Bcl-2
Cell cycle control p53, p21, p27, p16, cyclins D1, E
Adhesion E-cadherin, beta-catenin, CD44, CD24, Claudin-1
Migration/motility CXCR4, alpha6beta4 integrin, Net1, matrix metalloproteinases
Angiogenesis VEGF, Flt-4, HIF-1alpha, pericyte markers
Immune responses CD68, CD45RO, CD3zeta, CD4, CD8, PD-L1, FOXP3, CD1a, cytokines
Inflammation NF-kappaB, COX2, CSF-1R
Fibroblasts Fibroblast activation protein-alpha, PDGF-beta

Fluorescence-based imaging technologies have been applied to models of cancer and patient samples for many years. The applications have spanned the range from in vitro studies in single cells, populations of cells, mixed cell populations, three-dimensional (3D) tumor models, and pathology of patient tumor samples, as well as imaging cells within mouse tumor models, including antitumor immune responses, dynamics of cancer growth and invasion, tumor angiogenesis and regression, and tumor cell movements. Furthermore, applications in drug discovery have been performed in cells and in small experimental organisms including yeast, Caenorhabditis elegans , Drosophila , and zebrafish, as well as monitoring tumors in rodent models with whole-body imaging of small mammals.

The present chapter focuses on the application of high-content analysis (HCA) to populations of cells, more complex tumor models, and in vitro and patient samples where large image datasets can be created and explored with computational and systems biology tools to create a bridge between cancer cell biology and tissue-based diagnostics. The investigation and integration of the continuum of single cells, cell populations, 3D tumor models, and patient samples is needed to define the molecular basis of cancer.

HCA, originally termed high-content screening (HCS), is a platform technology created in the 1990s to automatically image, analyze, store, and mine large image datasets based primarily on fluorescence imaging microscopy, although transmitted light is an option. HCA harnesses advances in automation of microscopy, image processing, image analysis, fluorescence-based reagents, automation of sample preparation, and relational databases ( Figure 25-2 ). There have been numerous books and reviews on the applications of HCA in basic biomedical research, drug discovery/development, and diagnostics. A broad range of fluorescence-based reagents for both live cell, kinetic studies and fixed-endpoint investigations have also been reviewed in detail. The major types of reagents, readouts, and selected on-line databases are listed in Table 25-2 .

Table 25-2 includes citations for References .

Extensive lists of additional reagent sources can be found online and in published catalogues.

Figure 25-2, The components of high-content analysis (HCA)

Table 25-2
Classes of Fluorescence-Based Reagents, Readouts, and Online Resources for HCA
Resource Type Application References
Fluorescent probe classes Chemical fluorophores Wide spectral range, easily attached to targeting molecules, some are environmentally sensitive, many useful properties
Nanocrystals Stable, bright, single excitation, narrow emission, best for multiplexing in fixed cells or cell surface markers
Fluorescent proteins (FPs) Multiple wavelengths, transient or stable expression, linked to targets, some are environmentally sensitive, photoactivated (or switched), live or fixed cell assays
Cellular labeling approaches Antibodies Target expression level and localization
FISH probes DNA copy number variants, RNA expression, including micro-RNA
FPs Target expression level, localization and dynamics, photobleaching or photoactivation for transport within or between compartments
Environment-sensitive probes Ion concentrations, membrane potential, hydrophobic compartments
Proximity probes FRET, colocalization
Enzyme activity Fluorogenic substrates, cleavable linkers
Organelle specific Nucleus (DNA), mitochondria, lysosomes, neutral fat, endoplasmic reticulum, etc.
Fluorescent biosensors FPs or combinations of FPs engineered to report on activation of biomarkers or pathways
HCA readouts Intensity Relative concentration of target
Distribution Distribution and dynamics of molecular targets in cells
Colocalization Similarity or difference in the distribution of two or more labels
FRET Very sensitive determination of close proximity of two labels
Morphology Texture, size, or shape of cells or organelles, aggregation
Lifetime Local chemical environment
Polarization Molecular interactions (bound vs. free)
Cell tracking Motility, metastasis
Kinetics Measure of any or all readouts over time
Internet databases Spectral PubSpectra, Fluorophore.com , others
Targets The Human Protein Atlas, The Cell: An Image Library, The BioGRID Interaction Database
Cell lines The Cancer Cell Line Encyclopedia, Cell Line Navigator
FISH, Fluorescence in situ hybridization; FRET, fluorescence resonance energy transfer; HCA, high-content analysis.

Although the development of HCA has focused on the application of fluorescent probes, chromogenic probes continue to be used extensively for labeling tissue sections. Table 25-3 ∗∗

∗∗ Table 25-3 includes citations for References .

compares the advantages and disadvantages of fluorescent and chromogenic probes. Most HCA systems are optimized for the use of fluorescent probes, principally for their high sensitivity, high specificity, broad range of cellular functional readouts, broad range of wavelengths for multiplexing, and ability to engineer cells to express fluorescent proteins and biosensors. Because HCA makes use of automated imaging and quantitative image analysis, there is no need for direct viewing of the labeled specimen, and once the images are acquired, there is no further need for the specimen other than for institutional or clinical requirements. In traditional pathology, on the other hand, chromogenic probes have some advantages. The human brain is still the most sophisticated and reliable image processor for the interpretation of small numbers of images. Readily available, low-cost chromogenic probes provide stable and dense labeling for visualization in a transmitted light microscope or by digital image pathology, while simultaneously viewing the contextual morphology of the cells. Although providing somewhat lower resolution and more limited multiplexing than fluorescent probes, chromogenic probes still provide a good labeling strategy where one to three biomarkers per slide can be useful.

Table 25-3
Comparison of Fluorescent and Chromogenic Readouts
Reporter Type Advantages Disadvantages
Fluorescent Present standard in cell analysis
High sensitivity and specificity
Quantitative readout
Multiplex targets that are colocalized and/or in close proximity
Broad spectrum of wavelengths
Higher resolution with confocal imaging
Reagents are less stable for long-term storage
More expensive fluorescence, more expensive imaging systems
More expensive reagents
Chromogenic Present standard in tissue analysis
Long-term stability of labeling
Brightfield microscopes
Greater amplification
Variable sensitivity and specificity
Multiplexed targets must be spatially separated
Precipitates cause fuzziness around target

Success in the Human Genome Project demanded tools to define the functions of the coding and noncoding portions of the genome, to define the dynamic interplay of cellular constituents within and between cells, and to characterize subpopulations, as well as to define the relationships between populations of cells in higher order biological systems. The field was named cellomics, and the platform technology was named HCA . HCA harnesses the ability to implement combinatorial treatments on large sample sizes by using microplates, patterned microarrays and microfluidic devices for cells, microplates for small organisms, and mounted sections/microarrays for tissues. These large sample sizes are required for statistical analyses and exploration by computational and systems biology.

Imaging Live Cells and Model Organisms with HCA

Imaging living cells and model organisms by HCA has the advantage of allowing the investigation of the dynamic, temporal-spatial interplay of cellular constituents that define normal and abnormal cell and tissue functions. Single time points and/or kinetic measurements can be generated and analyzed. It is also possible to harness advanced, fluorescence-based probes and biosensors to measure physiological parameters not readily measured in fixed samples, such as cyclic protein translocations, pH, free Ca 2+ , membrane potentials, and a growing number of physiological biosensors by fluorescence microscopy. A disadvantage of investigating living systems by HCA is that biological processes can change from the time of imaging the first well in a microplate to the last well and this issue is multiplied when going from 96 to higher-density well plates. Depending on the time course for the specific biological process, including cyclic changes and the protocol for the addition of experimental treatments, large-scale, living samples usually have to be profiled in smaller batch sizes.

All HCA profiling or screening studies start with living systems that receive some combinatorial application of small molecules or biologics, RNAi for knockdowns, and/or nucleic acids for transfections or transductions. Although more demanding to perform, a recent investigation studied the kinetics of response of individual cells to drug treatments demonstrating the variability of cellular responses in a population. Measuring kinetic responses should increase as even more biosensors are developed and the complex and dynamic aspects of signaling processes are investigated.

Imaging Fixed Cells and Model Organisms with HCA

The main advantage of using fixed samples is that large-scale sample preparation and robotic screening of many microplates or slides is possible without changes in the biology during the readout. Therefore, many combinatorial treatments can be prepared at one time and the microplates/slides stacked in a robotic system for screening/profiling. There are many fluorescence-based reagents including antibodies, fluorescence in situ hybridization (FISH) probes, and fluorescent proteins that can be used to define single time-point localizations, relative concentrations, and activities. In order to optimally interpret fixed samples, either the half-time of a process under investigation must be determined in live sample profiles, or multiple time points must be generated in distinct wells or plates.

HCA has been extensively applied as a phenotypic approach to cancer drug discovery over the past few years, in both primary and secondary screens, either using live-cell or fixed-cell screening. Although specific molecular targets guide many of these screens, pathway modulations and phenotypic profiling are central to the approach. Examples of cancer biologies explored include energy metabolism, viral induction, apoptosis, cell cycle, autophagy, tumor invasion and metastasis, pathway modulations, a panel of biologies, and phenotypic changes compared to mutants. In many cases, HCA is also used in structure-activity relationship (SAR) to optimize lead compounds. However, it is still important to ultimately identify the mechanism(s) of action of lead compounds. The role of HCA in cancer drug discovery and development has been further advanced with the application of more quantitative analyses of profiles using computational biology and systems biology approaches, as explored in detail next.

Multiplexed to Hyperplexed Fluorescence-Based HCA

It has been the goal of imaging cytometry to increase the number of specific molecular parameters that can be measured in the same sample, so that complex interplays of components, pathway mapping, and heterogeneity of biological processes can be analyzed in increasing detail. We have defined multiplexed fluorescence in imaging applications as the combination in a sample of up to seven fluorescent probes that can be discriminated by spectral selection. Multiplexing has been accomplished in both live and fixed samples using a range of fluorescent probes. Multiplexing by flow cytometry has reached the level of 15 to 18 distinct fluorescent probes, but flow cytometry does not permit analyses of the temporal-spatial dynamics within or between cells. Hence, imaging technologies are being advanced to produce more parameters, especially in fixed samples. There have been a number of technical developments to increase the number of fluorescently labeled antibodies and FISH probes per sample, including new types of probes such as quantum dots, new algorithms such as spectral unmixing, and new protocols such as sequentially labeling, imaging, and quenching the fluorescence, and then repeating the process. Recently, the GE Global Research Center has demonstrated that more than 60 fluorescence-based biomarkers can be applied to a single tissue sample using a sequential labeling approach. This novel platform technology should have a great impact on basic cancer research, drug discovery, and diagnostics/prognostics. Generating the multiplexed to hyperplexed datasets creates a powerful platform that will enable the application of advanced computational methods to directly define pathways and modifications due to perturbations, as well as to characterize and understand heterogeneity. It is also possible to harness fluorescence lifetime imaging to gain some parameters, as well as the application of mass spectroscopy applied to single cells and tissues, but these latter approaches are not covered here.

In addition, other imaging modalities have been applied to cancer model systems and patients. The data from these investigations must be integrated into the systems biology models developed in cells, small experimental organisms, and patient tissue sections.

Cellular Systems Biology in Cancer Research and Drug Discovery

You're Reading a Preview

Become a Clinical Tree membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here