June 1, 2005 (Vol. 25, No. 11)
Proprietary Platforms Slowly Move Forward
Many companies are shifting their focus from genome mapping and sequencing to determining gene function. Functional genomics uses a range of technologies for genome-wide analyses supported by data interpretation. These activities depend on experimental and computation methods.
A Parallel Approach
Realizing that the application of overexpression for discovering potential targets has several disadvantages (off-target effects, false positives, toxicity), Xantos Biomedicine (www.xantos.de) now incorporates RNAi as a tool to validate targets.
“It’s a great way to use overexpression and knock-down in parallel to answer different questions regarding target validation,” says Angelika Bonin-Debs, Ph.D., director of business development and strategic alliances.
The company has a proprietary high throughput gene-transfection and assay platform (analyzes up to 300,000 cells/month), called XantoScreen, that is used for several function applications (molecular drug profiling, target/pathway profiling and identification, and lead identification). cDNA screening tools include a human clone collection of 35,000 full-length clones and several cDNA libraries from various human tissues.
“SiRNA is being used mostly in our target validation program, especially in the oncology area,” Dr. Bonin-Debs explains. “If you have a target with some significance, and you want to knock it down in a cancer cell, you can show the phenotype is no longer present.
Dr. Bonin-Debs says the company is looking to further expand its platform using this dual-approach. “We would like to combine the two approaches more and be able to use them in parallel where we look at what happens when a gene is overexpressed versus what happens when it’s knocked-down.”
MicroRNAs Linked to Disease
“A few years ago, Ambion (www.ambion. com) sensed that microRNAs (miRNAs) were an important component of biology and that their mis-regulation and/or mutation led to disease,” says David Brown, Ph.D., associate director of R&D. In-house studies of miRNA expression in cancer samples have identified that several of these molecules are differentially regulated in tumors.
“We found about seven to eight miRNAs that are almost always differentially expressed in tumors and are contributing to disease. We suspect many of these are regulating cell functions.”
Ambion launched its first miRNA products about 18 months ago. Its initial goal was to analyze miRNA expression because standard RNAi isolation procedures do not recover miRNAs. The mirVana microarray platform contains the mirVana miRNA isolation kit, labeling kit, and the mirVana miRNA probe set.
This platform enables parallel expression of miRNAs in human and mouse studies. An isolation process quantitatively recovers miRNAs from tissue or cells and purifies it from other RNA molecules. The miRNAs are then fluorescently labeled for microarray analysis, which can simultaneously assess the expression profiles of 200+ miRNAs.
Array data is confirmed by transfecting cells with molecules that can up- and down-regulate each known miRNAidentifying ones that are involved in various cell processes. “We like to combine miRNA expression data and miRNA functional data to determine if a gene might be important in disease,” states Dr. Brown.
“Right now people are projecting that there are between 500 to 1,000 unique miRNAs, so I would not be surprised to find that 90 to 95 percent of all human genes are regulated by them.”
High Throughput siRNA Synthesis
Qiagen’s (www.qiagen.com) siRNA design algorithm ensures greater specificity of gene silencing. “Our aim is to move gene silencing from experimental to a tool,” explains Eric Lader, Ph.D., associate director, gene silencing.
Licensed from Novartis, the algorithm uses neural network technology that predicts potent siRNA sequences based on a large database of experimental results. These siRNAs have high specificity and flexible design for any gene, species, splice variants, and specific mRNA regions.
A proprietary homology analysis tool used with the algorithm reduces off-target effects. It detects and rejects siRNA that knocks down an unintended target. “With siRNA, there are two main questions: are you missing important things and are you picking up artifacts? We are addressing both with our HiPerformance siRNA design,” states Dr. Lader.
There are several advantages to this method. It eliminates time-consuming siRNA design and optimization steps, offers high specificity (at least two of the four siRNAs will knock down the target), and has a wide range (up to 10 siRNAs per target are designed using the algorithm, and they are available for any accession number).
Qiagen offers siRNAs targeted against gene families in major drug target classes, including GPCR, kinase, cancer, phosphatase, and apoptosis genes. Dr. Lader says that the company plans to do more custom screening and validation procedures as well as develop methods to make siRNA delivery more specific.
Computational Tool Detects Disease-Related Genes
Researchers at Roche Palo Alto (paloalto. roche.com) have developed a computational method that quickly detects disease-causing genes. The pattern of genetic variation in the genome of different strains is analyzed using in-bred mouse models.
“We map and compare the pattern of observed differences (physiological/pathophysiological) among strains and ask where do these differences best correlate with patterns of genetic variation in the genome? Something that used to take many years, can now be done in an afternoon,” states Gary Peltz, M.D., Ph.D., head of genetics and genomics.
Conventional methods for genetic analysis in mouse models usually require about five years because the mice have to be experimentally generated, then characterized and genotyped. This new method accelerates the process.
“We began a few years ago to analyze the pattern of genetic variation among mouse strains and put that into a publicly available database. Then we developed a proprietary algorithm to do that type of comparison.”
According to Dr. Peltz, this method is currently limited by the amount of genome covered where the pattern of genetic variation is known (currently 10 to 15%) and the number of mouse strains analyzed (about 18). However, he says they hope to have a much larger part of the mouse genome analyzed and have 40 strains well-characterized within two to three years.
“Once these limitations are removed, there are many areas of biology that it is applicable to besides pharmacogenetics, gene expression, and transcriptional analysis. We’re working with a number of different therapy areas to analyze traits relevant to different diseases,” adds Dr. Peltz.
Analyzing Mechanism of Action
Through a former alliance with Exelixis Pharmaceuticals, scientists at Bristol-Myers Squibb (www.bms.com) acquired compounds with beneficial therapeutic effects, but no known target.
The goal was to determine if these compounds would generate phenotypes in model systemsa process called “reverse chemical genetics.” If they did, says Petra Ross-MacDonald, Ph.D., senior research investigator, applied genomics, the group used both forward and reverse genetics to modify the phenotype.
Focusing on which genetic changes modified a compound’s effect provided clues as to what the compounds were targeting. The company reported two small molecule successes using this approach: a second target for some oncology compounds that could itself be an oncology drug target and a compound that hit signaling through GPCR, but at a novel point.
Although reverse chemical genetics already exists (testing large compound libraries) and has generated data on the effects of compounds in cellular assays, Dr. Ross-MacDonald says that for this to be useful in drug discovery, “we need to know their molecular targets, the proteins they bind.”
Also, functional genomics efforts are generating data on the role of particular proteins in cellular assaysand for this to be useful, “we need to know the effects of hitting those proteins with compounds.”
“There is a feeling that running the two types of assays in parallel would be a start,” adds Dr. Ross-McDonald. Although there is currently no high throughput technology to connect the two types of information, the company says it is looking at areas where it could be advantageous to run compounds and genes in the same assays.
Genome-Wide siRNAs Screening
Novartis Institute for Biomedical Research (www.novartis.com) is using its large-scale siRNA screening to identify genes involved in cell-signaling pathways, as well as identify novel genes also involved in key regulatory pathways.
Starting several years ago, identifying gene functions via high throughput cell-based assays, the company established an infrastructure, which was later applied to loss of function (siRNA/RNAi). “RNAi became popular for the ability to do screens that were previously only possible with model organisms in an unbiased fashion to identify genes that would have a certain function,” states Craig Mickanin, Ph.D., research investigator, functional genomics.
The company has designed a large library (via a collaboration with Qiagen) to conduct genome-wide screens in mammalian cells. “We have now made an siRNA collection that covers the entire human genome and represents two siRNAs targeting approximately 24,000 human genes. We use this to identify the potential function of genes in a number of cell-based models of pathophysiological processes.”
Dr. Mickanin says that his department is working to focus on drug discovery from a pathway approach. “The technology platform we have developed fits well into that paradigm where we can execute screenings aimed at identifying key members of these known signaling pathways to expand the repertoire of potential druggable targets.”
A Three-Hybrid System Approach
GPC Biotech (www.gpc-biotech.com) has developed a three-hybrid based approach for identifying and characterizing the interaction of small molecules with proteins. Nikolai Kley, Ph.D., vp, research, explains that the underlying principle for developing this system was to understand what small molecules really do, how they act, and to identify targets that may curtail the efficacy of compounds.
“This is important in kinase inhibitors and targets that may underlie potential side effects of a particular compound,” he says.
The three-hybrid system is an adaptation of the yeast two-hybrid system that has been widely used for the past 15 years for large-scale, protein-protein interaction studies using cDNA libraries. Two proteins come together and, when they interact, a signal is transmitted that makes the yeast cells grow under certain conditions and create proteins.
“Our system is based on this particular complex, but here the interaction of two proteins is not direct, but mediated by a small molecule. The activation of the signal that allows the yeast cells to grow is dependent on the small molecule. Using this approach, we can do large-scale screens of cDNA libraries, and in an unbiased fashion, identify proteins that interact with small molecules,” Dr. Kley explains.
This approach is now being adapted to mammalian cells, where it allows the detection of interactions that are difficult to detect in yeast. It also has the potential, says Dr. Kley, to be used for biomarker discovery.
Custom DNA MicroArrays
Progenika Biopharma (www.progenika.com) developed a technology platform for producing and processing customized microarrays.
These arrays are designed and printed in-house. The company says the platform is flexible. Each array produced tests only the specific subset of genes or genetic mutations that are relevant to the application in question.
Progenika’s Rat Genome 5K microarray consists of oligonucleotides specific for 5,353 nonredundant rat genes. Probes incorporated cover genes involved in a broad range of biological processes, with a specific emphasis on genes involved in metabolism and clearance of drugs and toxins. Experiments performed with a minimum number of animals can provide insight into mechanism of action of drug candidates and toxicological profiling in response to alternative molecular variants.
Progenika also designs specific probes incorporated into microarrays for polymorphism detection. The company’s recently released LipoChip detects mutations causing familial hypercholesterolemia, and carries probes to detect 189 individual mutations that are known to be relatively prevalent in the Spanish population.
Another microarray is being developed for the diagnosis of inflammatory bowel disease that currently analyzes 46 mutations (more will be added when they are discovered). The company expects to begin clinical trials this month.
