December 1, 2005 (Vol. 25, No. 21)
Automated Image Cytometry Is Just One of the New Tools Available to Scientists
As tools and techniques continue to improve, fluorescence and bioluminescence are playing increasingly important roles in assay development, the discovery and characterization of targets and leads, and high-content molecular and cellular pathway research.
Experts highlighted recent advances in this exciting field at Cambridge Healthtech Institute’s “Fluorescent Proteins for Cell Imaging and Drug Development” conference, recently held in La Jolla, CA.
According to Jeffrey H. Price, M.D., Ph.D., associate professor at The Burnham Institute, and CEO of Vala Sciences (www.valasciences.com), the “big new tool” in fluorescence microscopy is automated image cytometry, also known as high-content screening. New high-throughput microscopes automatically collect in-focus images of individual cells in microtiter plates, and automated image software analyzes the fluorescent micrographs.
Automated image cytometry allows researchers to conduct more experiments and collect “more and more information” from each project.
The technology can also improve data analysis by providing researchers with objective measurements and statistically significant data. “There’s less argument,” explained Dr. Price, because “instead of looking at images qualitatively, you get quantitative results, like IC50 and EC50 dose-response points.”
Because the instruments “can measure dozens of parameters per cell for tens to hundreds of cells per well, the information density is much higher than conventional HTS instruments, which are capable of only a few measurements per well,” Dr. Price noted. “The challenge is to analyze the information.” Since the field is so new, technology users are still learning the advantages and pitfalls of the measurements they can make, he added.
Dr. Price pointed to the work of another presenter at the conference, David Zacharias, Ph.D., assistant professor, Neuroscience, University of Florida, College of Medicine, as a dramatic example of the technology’s power. Dr. Zacharias used automated image cytometry to study the subcellular distribution of lipid-modified, fluorescent proteins.
“The fluorescence moves from cytoplasm to membrane but there is no change in intensity, so the plate reader can’t measure the change” to provide useful data interpretation, said Dr. Price. To automate the analyses, image analysis is required or a special new biosensor must be developed.
“Vala’s membrane analysis software was built using Dr. Zacharias’ images and experiments as the first biological application,” explained Dr. Price. The software, which is platform-independent, was then “expanded to other kits,” such as PKCa, N-cadherin, E-cadherin, and VE-cadherin. The combination of kits and software enabled high-throughput imaged-based assays of membrane translocation.
The range of Vala’s Thora image cytometry software is illustrated by the figure showing the performance of PKCa, N-cadherin, and E-cadherin high content screening (HCS) assays.
BioImage (www.bioimage.com) also creates cell-based assays that use fluorescence to monitor protein translocation. BioImage can make assays for “any protein in the cell that translocates within the cell in response to a pharmacological stimulus in a reproducible fashion,” said Len Pagliaro, Ph.D., vp, business development.
The protein of interest is fused to GFP, the plasmid construct is transfected into cells, and a stable cell line is generated. Although “the idea is very simple,” Dr. Pagliaro stressed that it is difficult to create a fully validated drug screening assay. “We make a big distinction between a cell line and an assay.”
To become validated assays, cell lines must demonstrate consistent protein expression levels and respond appropriately to stimuli over time. “For every assay you develop, you need to have a reference compound,” usually a small molecule that is used to confirm the biology of the cells, explained Dr. Pagliaro.
BioImage also uses RNAi technology to characterize its cell lines. RNAi “can provide excellent controls that users of high-content assays increasingly want to see,” Dr. Pagliaro noted.
With a validated assay in hand, drug screening and profiling begins. Translocation-based screens can produce traditional drugs like kinase inhibitors. “Much more interesting,” noted Dr. Pagliaro, is that these screens can also yield new compound classes that have “potentially better selectivity and novel modes of action.”
BioImage identified one compound series as a moderate potency hit in a Forkhead oncology screen. Although its exact molecular mode of action is not yet known, BioImage has performed lead optimization and SAR and have some “impressive xenograft data,” according to Dr. Pagliaro. The company is currently seeking partners to carry the project to an IND.
Daniel R. Rines, institute fellow, lead discovery, Genomics Institute of the Novartis Research Foundation (web.gnf.org/index. shtml), used automated fluorescence to conduct high-throughput screens for cell cycle and cancer targets. Rines’ goal was to “identify new genes that would cause cells to arrest in metaphase.”
A library of double-stranded siRNA was transfected into HeLa cells and U20S cells. The cells were fixed and stained with a fluorescently-labeled antibody against phospho-Histone H3 (p-His), an excellent marker for mitosis.
Since “histones are phosphorylated when chromosomes are being condensed, any cells stained positive for phospho-Histone H3 suggests they are in metaphase or G2/M,” explained Rines. The percentage of cells staining positive for p-His is an indicator of the mitotic index.
In a normal cell population, about 5% of the cells are in metaphase. But if a specific siRNA inhibits a gene that allows progression through metaphase, more cells stain positive for p-His.
A key advantage of automated fluorescence microscopy for HTS with microtiter plates is that you “can do single-cell statistics because the microscope can distinguish individual cells in the well,” said Rines.
The instrument calculated the total number of stained cells and the percent p-His positive, as well as data on cell morphology, such as the size of each nucleus, its brightness, and the cell cycle stage. On average, 2,000 cells were analyzed in each well.
Rines performed the genome-wide screen and found “six different phenotypes from eight novel genes,” all involving defects in the mitotic spindle. Since the screen collected so much data, including morphological properties, such as the number of nuclei per cell, Rines was also able to identify cytokinesis genes in addition to those affecting the cell cycle and progression through metaphase.
“Because it’s all based on cell morphology and sub-cellular protein localization, you can ask not only multiple questions in a single screen but the kind of questions that can’t be addressed by any other technology. Standard high-throughput screens typically only ask one question, such as whether a protein is expressed or not,” said Rines. That’s what “makes this approach very powerful.”
Frank Wunder, Institute of Cardiovascular Research, Bayer HealthCare (www.bayer.com), outlined a novel approach for identifying modulators of the cGMP pathway. Cyclic GMP “is a central mediator of vasorelaxation” so modulators may be beneficial for the control of blood pressure and hypertension.
Traditional assays for cGMP are labor-intensive and expensive. “We have a library of over 1million compounds, so we cannot perform the HTS by radioimmunoassay,” Wunder said. To overcome this challenge, a bioluminescence-based cell line was created to enable high-throughput screening for cGMP modulators. The cells constitutively expressed soluble guanylate cyclase (a1/b1-heterodimer). When activated by a soluble guanylate cyclase activator or NO, soluble guanylate cyclase converts GTP to cGMP.
The cGMP then activates the cation channel CNGA2, which is also constitutively expressed in the cell and is selectively activated by cGMP. When activated, this channel conducts calcium ions from the extracellular medium into the cell.
Wunder’s team considered two possibilities to detect incoming calcium, calcium-sensitive fluorescent dyes or aequorin, a calcium-sensitive photoprotein, but ultimately chose to use aequorin.
“Aequorin has at least one advantage over fluorescent dyes. If you do HTS and use fluorescent dyes, you always would have to look for fluorescence interference from library compounds you are screening. You can circumvent this problem if you use bioluminescence,” explained Wunder.
More than 900,000 compounds were screened, resulting in 500 confirmed hits. Known stimulator classes like NO donors, organic nitrates, and the BAY 41-2272 class were confirmed. A new class of guanylate cyclase activator, BAY W1449, also emerged from the screen. BAY 58-2667, which was subsequently identified through a chemical-derivatization program, has since been pharmacologically characterized and is now in pre-clinical development.
John Shultz, Ph.D., group manager, R&D, Promega (www. promega.com), has worked to optimize and expand the usefulness of bioluminescence assays. These particular bioluminescence assays are based on three reaction components, luciferin, ATP, and luciferase, which when combined, produce light.
“In general, bioluminescence assays are much more sensitive than fluorescence in terms of detection limit because the background is very close to zero,” Dr. Shultz said.
For many years, Promega only sold methods to measure two of the three bioluminescence reaction components, luciferase and ATP. To extend the flexibility of the assays, Promega recently developed reagents to allow the measurement of luciferin generated in reactions where a second enzyme transforms a luciferin precursor into a form that can be utilized by luciferase.
In this way, the generation of light can be used to indirectly follow the activity of other enzymes. For example, Promega researchers have engineered a version of luciferin that is linked to a short peptide, called DEVD-luciferin.
The enzyme caspase cleaves this precursor into luciferin. This caspase assay “can measure enzyme concentration over four to five orders of magnitude of activity because it’s working with such a low background,” said Dr. Shultz. “The assay has excellent linearity, and the signal generated by the system is extremely stable over time.”
Promega also created a number of precursor compounds that are selectively metabolized into luciferin by specific isozymes of the major Cytochrome P450 enzymes, which are involved in drug metabolism and often studied in ADME/Tox investigations.
The assays can be used both to measure the activity of the enzyme itself and to accurately measure the effectiveness of compounds that inhibit the enzymes. Similar assays have also been developed to measure MAO and GST activity.
Tamara Troy, senior scientist, Xenogen (www.xenogen.com), quantitatively compared standard bioluminescent and fluorescent reporters used for in vivo imaging. “Given the wide use of reporters, there’s a need for quantitative comparison,” said Troy.
In the models Troy examined, the fluorescent signals were brighter but bioluminescent reporters were more sensitive because of a higher signal to background ratio.
“The biggest limitation with fluorescent imaging is the presence of tissue autofluorescence,” Troy said. To limit autofluorescence and background fluorescence, Troy suggested feeding mice alfalfa-free diets, choosing fluorescent markers in the near infrared, and using techniques like background subtraction or spectral unmixing.
One of the limitations of in vivo optical imaging has been the precise determination of the location of the source of the signal. “Single-view, single-wavelength images do not constrain the source to a unique solution. A collection of small, shallow sources looks similar to a deeper, bigger source because of tissue scatter and absorption,” explained Troy.
To address this issue, Xenogen developed new 3-D localization reconstruction software for use in bioluminescence imaging, known as DLIT, released earlier this year. The company is currently working on similar technology for fluorescence imaging. The software relies on surface tomography information and multiple wavelength data (images acquired with filters ranging from 560660 nm) to reconstruct source location and brightness.
“If you have a deep, bright source it will be red-shifted because shorter wavelengths will be absorbed and longer wavelengths will be attenuated. If the signal is red-shifted, you know it’s deeper,” explained Troy. “Multiple wavelength data helps constrain the source location to a unique solution.”