April 1, 2005 (Vol. 25, No. 7)
Recent Advancements Could Impact How Researchers Decipher the Role Proteins Play
In the quest to unravel the mysterious behavior of proteins, scientists gathered to exchange information at the first annual “U.S. Human Proteome Organization (HUPO) Symposium,” held March 13-16 in Washington, D.C. The topics ranged from protein profiling and mapping to sample-preparation methods, to bioinformatics. This article takes a look at how these new advances will impact proteomics.
New Tools and Reagents
Researchers from Beckman Coulter (www.beckman.com) presented a two-dimensional approach using a capillary tube for peptide mapping that accommodates target protein samples in femtomol quantities. By minimizing the steps between digestion and mass spectrometry (sample cleaning, desalting, and concentration), the sample integrity is maintained.
“This is a new approach for capillary electrophoresis. Part of the main challenge is detection. The benefit of our approach is that the sample injection volume is no longer limited to the capillary dimensions. We can use large volumes (microliters), which translates into good concentration sensitivity,” stated Chitra Ratnayake, Ph.D., team leader, proteome lab development group.
The 50-micron diameter capillary tube has a proprietary solgel plug at the inlet end that binds peptides. Different types of appropriately modified particles can be used in the same solgel concentrator to selectively concentrate analytes of interest.
The concentrated sample is then separated into individual components. Accurate protein characterization can be achieved using a sample as low as 50 femtomols in less than one hour. “This approach allows us to analyze low abundance proteins automatically and more rapidly,” added Dr. Ratnayake.
Applied Biosystems (www. appliedbiosystems.com) demonstrated the capabilities of its iTRAQ reagents (amine-specific isobaric reagents), which can reduce, digest, and label all peptides (including post-translational modifications) in a single tube, eliminating potential sample loss.
All reagent-labeled samples are then combined into one sample for LC-MS/MS analysis. Up to four samples can be prepared and analyzed simultaneously. This provides a simple workflow, which has been shown to be ideal for biomarker research, including discovery, identification, and quantitation (relative and absolute).
“We’re focused more on discovery and early validation of biomarkers,” said Sally Webb, proteomics marketing manager for the Americas. “In the past, the main approach has been profiling. Now it’s important to identify potential markers because we have to know what they are in order to give them a function and know they are biologically significant.
“These reagents provide simultaneous quantitation and identification so you know whether these markers play a role in the biological system.”
iTRAQ reagents can be used to discover biomarkers in tissue samples or biological fluids. In addition, the reagents were used in early validation work, providing absolute quantitation of expression changes of cytochrome P450 isoforms with drug treatment.
RNAi and Protein Quantitation
A new way to combine RNAi knockdown technology with protein quantitation methods (isotopic labeling with peptides and 18O water) was presented by Sigma-Aldrich Biotechnology (www.sigmaaldrich.com).
“These isotopic peptides may be the only way to absolutely quantitate the amount of protein you have in your system of interest,” said Graham Scott, Ph.D., R&D manager, protein expression and proteomics.
The process involves digesting a protein sample, choosing one or two peptides, and then making heavy synthetic analogs of those by inserting an amino acid with stable isotopes (Carbon-13 or Nitrogen-15).
“Since you know exactly how much heavy analog you added, when you run your MS, the heavy peptide will show up at a slightly different point relative to the native peptide, but close enough for a direct comparison,” explained Dr. Scott.
Traditionally, researchers have used Western blotting to assess how much protein expression was knocked-down with RNAi. However, Dr. Scott said, this technique is not very precise, and that “with the isotopic peptides, you can interrogate any protein in theory; simply by synthesizing a relevant peptide.”
The 18O water method requires a control sample’s (no RNAi added) peptides to be labeled with 16O (normal oxygen) and an affected sample’s peptides (RNAi added) labeled with 18O (heavy oxygen). Equal amounts of the control and affected samples are then mixed and run on a mass spectrometer. The relative heights of the two peaks are compared.
“The advantage of this method,” said Dr. Scott, “is that it’s a global strategyyou can see the relative differences between all peptides in the sample.”
2-D gel performance is influenced by several factors, including gel size and staining techniques. Although the 2-D gel approach is considered by some to have low reproducibility, and to be slower and more cumbersome than the LC approach, “it produces very close to single protein spots, which can then be excised,” stated Aran Paulus, Ph.D., R&D manager, new technologies, expression proteomics, Bio-Rad Laboratories (www.bio-rad.com).
Enhancing 2-D Gel Analysis
Dr. Paulus and colleagues analyzed two different gel sizes: 11 cm and 24 cm. The 11-cm gels are faster (12 days run time) and are best for a quick overview of your protein, said Dr. Paulus, whereas the larger gels (~4 days run time) allow more sample volume, provide a higher spot count, and can dig deeper into the proteome. “If you need to look at more low abundant proteins, you need the larger gels,” he explained.
The group also compared three stains in terms of practicality, price, and sensitivity. Coomassie Blue is very easy to use, inexpensive, can be viewed with the naked eye, but isn’t very sensitive. Silver is more sensitive, but development is tricky and it is not mass spectrometry compatible.
Sypro Ruby requires a fluorescent scanner and is the most expensive stain, but one, Dr. Paulus said, that is the best for biomarker discovery, because of its high sensitivity, high dynamic range (about four orders of magnitude), and ease of use.
“You have to adjust the loading and dynamic range of your dye to the level where you want to see your proteins of interest,” he summarized.
Better Protein Profiling
Waters (www.waters.com) has developed a platform for the simultaneous qualitative and quantitative protein profiling in a single LC/MS run without isotopic labeling. The Protein Expression System incorporates a proprietary UPLC (nanoACQUITY UPLC) with a proprietary mass spectrometer (Q-Tof Premier), and is programmed to acquire data with a dual-function exact mass protocol.
“We collect alternating high and low collision energy data under exact mass conditions,” explained Tim Riley, Ph.D., vp, proteomics business development. This differs from other systems that use data dependent-directed LC-MS/MS.
“You lose a lot of information, because when conventional systems are collecting MS-MS, other peptides are eluting. So, it really compromises the quality of the quantitative information that might be contained in the low-energy information set,” said Dr. Riley.
Since low- and high-energy data from each component in the sample is obtained, Dr. Riley said, “we get improved qualitative characterization because we’re getting MS-MS on just about everything, instead of maybe 2030% of all the peptides in the mixture.
At the same time, in the same analysis, we can use all the low- energy information to quantitatively characterize the magnitude of abundance of each peptide in the sample.” Proprietary algorithms match the like peptides between two sample injections, and are based on the accurate mass and retention time signatures of each peptide.
“Our protein expression system allows users to do an enhanced qualitative analysis and simultaneously gather relative quantitative information. It’s important to see what has changed between two sample mixtures when looking for biomarkers,” Dr. Riley summarized.
Focusing on Sample Prep
Companies are beginning to focus more on methods to enhance sample preparation and integrity. In order to better analyze low-abundant proteins for potential biomarker discovery, researchers at Agilent Technologies (www.
agilent.com) have developed a new, high-capacity HPLC column.
Based on a novel attachment process for affinity binders, the column removes the six most prevalent proteins from human serum. “There are extreme challenges in analyzing serum; protein concentration is in the range of 10 to 12 orders of magnitude. To reach the biomarkers, you really need to dig deeper by depleting the high abundant proteins,” explained Nina Zolatarjova, Ph.D., R&D scientist.
This high-capacity column is an upgraded version of a column introduced by the company last year. Users can now add 4050 L in the 4.6 X 50 mm column and 80100 L in the 4.6 X 100 mm column.
“This column is more robust than our previous version. We ran it 300 times without any loss of specificity or capacity,” said Dr. Zolatarjova. The main application will be reduction of sample complexity for analysis of control versus disease samples using 2-D gels and LC-MS. The company anticipates the column will be available this month, along with a spin-tube version for labs that don’t have HPLC capabilities.
BD Diagnostics (www.bd.com) has been researching the instability of plasma proteins in post-collection samples and is developing a chemistry to address some of the issues of protein degradation. “Once plasma proteins are pulled from their protective’ in vivo environment, ex vivo losses start within milliseconds, within a few minutes or hours, results may be unobtainable.
“We used MS to find proteins that appear to be damaged over time (up to 72 hours post-blood draw), and demonstrated a very basic first look that we can eliminate that damage,” said Bruce Haywood, business development manager, proteomics.
Quantitative analysis indicated relative intensity of peptides changing with time, suggesting that some plasma proteins are being digested by intrinsic proteases.
A protease inhibitor cocktail included in blood collection tubes at the time of sample acquisition stabilizes the peptide mass spectra over time. Protease activity may obscure downstream analyses by introducing variability that can mask the value of protein biomarkers.
The researchers concluded that their studies indicate “protease inhibitors provide a net benefit in preserving a more in vivo-like plasma sample, providing potentially more robust samples for biomarker discovery.”
Biomarker Discovery Platform
Predictive Diagnostics (www. predictivediagnostics.com) has combined several technologies into a biomarker discovery platform. This includes: an automated high throughput biomarker enrichment technique with novel membrane chromatography, high-resolution detection via MALDI Orthogonal-TOF mass spectrometry, and biomarker analysis via the company’s BAMF (Biomarker Amplification Filter) algorithms.
This platform has been used to identify biomarkers for Alzheimer’s disease and lung, prostate, and ovarian cancers. “We start with a normal and a disease blood sample, run these on mass spectrometry and then input the raw data into our BAMF technology processors,” Guy della Cioppa, Ph.D., senior vp, business development, explained.
Control spectra versus disease spectra is aligned and analyzed for differences. “We typically look at the low molecular weight proteins and have developed some proprietary ways to enrich the relevant biomarkers in that weight range,” he added.
The BAMF platform runs on several large Linnux clusters, essentially parallel supercomputing. “We figured out some proprietary ways to reduce model space so we can integrate these very large marker model sets and build an optimal diagnostic fingerprint.”
A new, automated method for scanning LC-MS data sets for peptides and proteins that provides quantitative profiling and interactive confirmation was presented by GE Healthcare (www.gehealthcare.com). The DeCyder MS software provides a two-dimensional representation of LC-MS data.
“This shows mass spectrometrists things they haven’t seen before in their data versus conventional software,” says Harald Pettersen, Ph.D., scientist, discovery systems. For example, the 2-D representation allows the user to see contaminants and whether the chromatographic resolution is satisfactory.
“It’s much faster to look at the data in 2-D, where you can see all the data at once, instead of looking at small portions one at a time,” explained Dr. Pettersen. Instead of taking several days, the software provides analyses within an hour because the data is run automatically.
The 2-D image looks similar to a 2-D gel. Algorithms used to match images are adapted from the original DeCyder software (developed by Amersham Biosciences).
“The goal is to find peptides that are up- or downregulated between samples and to quantify them. We show that quantification using LC-MS data without using any chemistry is fully feasible. For researchers working with peptides, where you can’t use 2-D gels, this is unique,” summarized Dr. Pettersen. The company plans to launch the software in the second quarter of this year.
Correlation Network Analysis Identifies Biomarkers
A research group at the Johns Hopkins Center for Biomarker Discovery, in collaboration with Ciphergen (www.ciphergen.com), has developed a correlation network analysis to detect and visualize co-regulatory patterns in a large number of variables in a system.
“The idea behind this is that most people are doing comparisons between one gene or one protein expression across different phenotypes. But, in the case of cancer, there are multiple genes or proteins that work together in the disease process.
“We’re trying to understand how these things act together and how they relate,” states Zhen Zhang, M.D., associate director, Center for Biomarker Discovery, Johns Hopkins University School of Medicine (JHU).
Instead of analyzing samples that are too complicated (e.g., serum), the researchers found they had better results if they looked at one sub-protein at a time. These are analyzed via immunoprecipitation (IP) pull down by antibodies followed by Ciphergen’s SELDI data analysis.
Differential analysis of correlation networks using the SELDI data enabled the researchers to detect and later identify a marker that is upregulated for ovarian cancer.
The researchers also looked at genes differentially expressed in cultured ovarian tumor cells before and after drug treatment. “We wanted to see if anything translates into the protein expression levels,” says Daniel Chan, M.D., director, Center for Biomarker Discovery, JHU.
“Once you restrict the number of proteins you are looking for, you can use the correlation network to analyze the link between gene and proteins, and to analyze changes in a collective way that is directly related to disease.”