January 15, 2008 (Vol. 28, No. 2)
Verifying and Prioritizing Findings Aided by New Tools and Strategies
Although high-throughput biomarker research has given life scientists a huge amount of data with which to work, challenges remain for those carrying out studies in this area. A number of leading scientists will gather at Cambridge Healthtech’s “Molecular TriMedicine Conference” in San Francisco in March to talk about ways to advance biomarker research while simultaneously addressing key roadblocks to these research efforts.
Material for plasma biomarkers can be obtained in clinical practice through simple blood work, but discovering these low-abundance circulating biomarkers can prove difficult. Daniel Chelsky, Ph.D., CSO for Caprion Proteomics (www.caprion.com), is scheduled to discuss the company’s strategy of backtracking biomarker discovery to the Golgi where many potential plasma biomarkers originate and are most concentrated.
Dr. Chelsky has results from two pilot studies that captured known and previously unknown biomarkers from isolated and Golgi-marker verified samples. Caprion used its CellCarta® platform, a multistep process that includes mass spec detection and quantitative peptide expression profiling.
“We are always looking for proteins that are differentially expressed in a condition of interest such as in response to disease or a drug,” says Dr. Chelsky.
In the first pilot, researchers tested the hypothesis that secreted proteins from visceral adipose tissue, or belly fat, promote a suite of metabolic disorders including insulin resistance and glucose intolerance.
Out of three morbidly obese patients, Caprion found 155 proteins in Golgi samples, 65 known secreted proteins from other studies, and 83 additional secreted proteins as possible therapeutic targets.
Several of the known proteins were involved in lipid metabolism (lipoprotein lipase), signaling (prohibitin), cytokine/ growth factors (TGFB1), and metabolic processes (glucosidase 2 subunit B). Five potential targets are already in drug trials for various disorders by other companies.
Caprion is recruiting 30 more patients to extend its study to compare secreted proteins from visceral and subcutaneous fat, which may explain the connection between belly fat and metabolic disorders.
In a prostate pilot study, Caprion researchers identified 40 proteins known to be associated with cancer as well as 20 unexpected secreted proteins that distinguish healthy from cancerous tissues in the same patient. In order to prioritize discovered secreted proteins, a signal sequence search is conducted and proteins with known related biological function (such as cytokines) and signaling proteins with no obvious connection to the condition will be investigated. Pragmatically, proteins are also prioritized with off-the-shelf detection reagents when available.
Building Discovery Pathways
Biomarker discovery at the benchtop level can cull massive amounts of data, but putting that data in context is important to prioritize further research. Part of the solution is to know what research came before, but culling through thousands of pages of literature can be laborious.
Ingenuity Systems (www.ingenuity.com) claims that it has already solved the problem with its Ingenuity Pathways Analysis (IPA), a literature-based database originally developed for large datasets like transcriptomic arrays. “With an expanded version, we are now able to support chemical data as well as enable researchers to look for potential biomaker candidates within that data,” says Arthur Okamoto, Ph.D., senior product manager.
IPA can be used to analyze data from diverse sources including transcriptomic, proteomic, and metabolomic platforms side-by-side. “Being able to compare those three platforms would be incredibly powerful,” according to Dr. Okamoto, “but literature-based databases have been troubled with ontological ambiguities and conflicting interpretations.
“The best way to address ambiguity in the literature is to essentially use people reading articles,” continues Dr. Okamoto. In fact, Ingenuity Systems employs a number of Ph.D.-level scientists who curate the literature by hand. “We found the natural language processing or text-mining techniques aren’t able to resolve those ambiguous terms very well,” says Dr. Okamoto. “But when you have the ability to access the entire article as our curators do, you’re able to identify and specify the correct biological terms.”
Not only can proteins be targets for biomarker discovery, but their conformation can also be indicative of disease. Around 40 amyloid diseases are now known. These protein-associated disorders take on alternative conformations commonly noted in neurodegenerative diseases such as Alzheimer’s.
“These diseases are fairly significant in impact and yet there are few examples of efficacious drugs, even the acetylcolinase and esterase inhibitors, which are on the market. Most clinicians view these therapies as fairly marginal in their efficacy,” says Alan Rudolph, CEO of Adlyfe (www.adlyfe.com). “It’s a great opportunity for biomarkers to have an impact in a number of significant ways.”
Unlike many biomarkers, those against amyloid must not only detect the protein, but distinguish conformation, a unique challenge for biomarker discovery.
To overcome that challenge, Adlyfe developed a new assay. In it, a small sequence matching peptide that, when bound to the amyloid target, undergoes an alpha helix to beta sheet conformational change resulting in a positive fluorescent signal. Adlyfe’s technology measures beta-rich secondary structural forms of these proteins, “thought to be key markers of the earliest form of disease or signs of dementia,” according to Rudolph.
The nucleation of additional peptides amplifies the signal, a method Rudolph describes as “the PCR of conformational protein research.” The pronucleon ligand can cross the blood-brain barrier and can image amyloid aggregates in the brain, typical of neurodegenerative diseases.
Major challenges for biomarker discovery in the neurodegenerative field will be monitoring the estimated 350 antiamyloid drugs in the pipeline and detecting early stages of neurodegeneration. “Memory loss doesn’t necessarily mean you have Alzheimer’s,” says Rudolph.
Discovery to Validation
In the biomarker discovery pipeline, how researchers transition from discovery to verification, while maintaining high throughput, presents a challenge.
One possible solution: multiple reaction monitoring (MRM), a sensitive, quantitative mass spectrometry (MS) tool coupled with global internal standards that bypass synthetic protein standards, an approach under development from Christie L. Hunter, Ph.D., senior staff scientist at Applied Biosystems (www.appliedbiosystems.com).
In MRM, based on triple quadrupole MS platforms, the first analyzer transmits the peptide ion of interest, the peptide ion is fragmented in a collision cell, then the second analyzer transmits only one of the characteristic sequence ions of interest to the detector.
MRM experiments can be quantitative, but one of today’s challenges in peptide MRM analysis is how to maintain this quantitative accuracy across many patient samples and push the throughput envelope. Good internal standards enable this.
Typically, researchers create internal standards by synthesizing the peptide of interest with a stable isotope labeled amino acids. However, this can become prohibitively expensive in protein biomarker verification when internal standards for 50–100 potential proteins are required. Translation: potentially hundreds of synthetic stable isotoped labeled peptides will have to be made.
Dr. Hunter now has developed “more generic, more global internal standards using chemistry rather than making synthetic peptides.”
In the proof of principal experiment, Hunter focused on profiling proteins from 20 sets of twins. Digested protein from all 40 samples was pooled and labeled with a heavy label to create a global internal standard. The label under development is a derivative of Applied Biosystems’ iTRAQ reagents for MRM, called mTRAQ reagents, a pair of nonisotopic amine labeling reagents.
Light-labeled individual samples were spiked with pooled heavy labeled patient sample, and the light and heavy peptides from putative biomarkers were monitored by MRM, providing normalized peptide and protein ratios for every patient sample. MRMs can provide high quantitative reproducibility with coefficient of variance of <10 percent and with Dr. Hunter’s generic internal controls, coefficient of variances can often squeeze under five percent.
Dr. Hunter said that after this verification and prioritizing of discovered biomarkers, the next step could include absolute quantitation. “Throughput is definitely a challenge that we’re all facing in terms of how many samples and proteins can be monitored during verification and we are working to maximize the numbers of MRMs in a single run.” (There are now 600 MRMs in a single run, but Dr. Hunter hopes to double that number.)
In some ways, biomarker discovery is only as good as the breadth of biologically relevant samples a researcher or clinician has available. To bring more information to the sample end of biomarker discovery, Ena Wang, M.D., senior staff scientist in the infectious disease and immunogenetics (IDIS) section of the Clinical Center, NIH (www.nih.gov), under the supervision of Francesco Marincola, argues that RNA amplification, coupled with serial sampling, is essential to identify biomarkers to predict disease progression or reactions to therapy.
According to Dr. Wang, because tumor biology and human immunology are so complex, there is a need for real-time analysis of specific tumors in order to begin to look at disease and treatment dynamics.
Though excisional biopsies deliver a large quantity of material to study, follow- up and time series are impossible. Using minimally invasive biopsies, such as fine needle aspiration, Dr. Wang’s group can sample from the same tumor or diseased tissue, which allows follow up. The team concluded that patients could tolerate minimally invasive biopsies for paired analysis of the same lesions, which opens up the potential for studies into the natural history of lesions and their responses to therapy.
To put principal to practice, the IDIS group profiled basal cell carcinoma treated with Imiquimod. Comparing pre- and post-Imiquimod therapy, they identified 1,500 differentially expressed genes, but the addition of placebo-treated samples decreased that gene pool to 600 genes. “We’re looking at the markers identified not only by melanoma and basal carcinoma, but also across other disease types, so it’s kind of a validation system through other systems,” explains Dr. Wang.
Dr. Wang’s research also focuses on finding rejection biomarkers in carcinomas and also in infectious and autoimmune diseases. “We believe the rejection pathways are common, and they depend on the kind of setting,” says Dr. Wang.