Send to printer »

Feature Articles : May 15, 2012 (Vol. 32, No. 10)

Maximizing Mass Spec's Potential

  • Josh P. Roberts

No longer confined to the realm of specialized laboratories, the use of mass spectrometry (MS) has exploded of late. Researchers from a variety of disciplines contribute to making the analysis of ever smaller samples of macromolecules faster, easier, more streamlined, less error-prone, and less paradoxical.

At the “International Symposium on Microscale Bioseparations and Analysis” held earlier this year, scientists discussed their research into various tools and techniques used in combination with MS.

Hartmut Schlüter, Ph.D., professor at the University Medical Center Hamburg-Eppendorf, concentrates on mass spectrometric proteomics—for example, detecting, purifying, and identifying unknown proteases for which the catalytic reaction is known. He described a special form of two-dimensional displacement chromatography

using the Agilent Technologies HPLC-Chip, a miniaturized system designed to separate small amounts of peptides—down to femtomoles to attomoles—for analysis by mass spectrometry.

“In collaboration with Ulrich Tallarek from the University of Marburg and Agilent Technologies, we are trying to improve the HPLC-Chip technique,” he explained.

Driven by pulsed injections of a displacer molecule, samples pass through a cation exchange (CEX) column—the first dimension—on the way to the chip, which itself consists of two columns. The first is a reverse-phase trapping column where tryptic peptides are enriched and desalted. A valve—also present on the HPLC-Chip—is then switched for the peptides to be eluted to and separated by a second reverse-phase column. From there, fractions are ejected out the integrated nanospray emitter into the mass spectrometer via an ionization chamber.

Peptides compete with each other for binding to the CEX stationary phase during loading, with those of the highest affinity remaining nearest the column inlet. Yet instead of using a salt gradient to elute peptides from this first dimension, Dr. Schlüter’s team uses spermine to displace the peptides from the column. Spermine competes with the peptides for binding to the stationary phase, so rather than diluting and potentially mixing the fractions, they are focused and concentrated.

Displacement chromatography faces at least two challenges. First, it’s critical to adjust the amount of sample to the binding capacity of that sample to the column—yet determining what that binding capacity is can itself waste potentially precious sample. Second, continuous flow of displacer—typical for one-dimensional or off-line separations—in online chromatography doesn’t allow time for samples to separate on the second, reverse-phase, column.

They were able to overcome these challenges by using repeated pulsed injections of sample, slightly overloading the column. Using LC-MS to monitor everything that passed through the CEX column following each injection, they determined that the column was fully loaded for the first low affinity peptides eluted.

Similarly, using small, pulsed injections of spermine and monitoring the LC-MS chromatograms after each allows for a physical separation of the eluted fractions and optimization on the fly. There is a significant gain in resolution over that of a salt step gradient typically applied, but it comes with a price, Dr. Schlüter noted, in that pulsed displacement chromatography is a far more time-consuming endeavor.

Let’s Get Small

Charged molecules can also be separated by electrophoresis. This can be done about 100 times faster on microfabricated structures than in capillary formats, pointed out J. Michael Ramsey, Ph.D., professor of biomedical engineering at the University of North Carolina at Chapel Hill. Case in point: platforms such as the Agilent 2100 Bioanalyzer. “But it has been limited to using fluorescent tags basically to identify biopolymers—DNA and proteins.” Electrophoretic mobility by itself essentially indicates little more than a relative measure of size.

Dr. Ramsey, who was the scientific founder of Caliper Technologies (now part of PerkinElmer), has been working on integrating a microfluidic capillary electrophoresis (CE) platform with electrospray (ES) ionization.

“So you can do these same types of high-performance separations that now get the biomolecules out into the gas phase so you can do MS on them,” he said, and allow chemical information such as molecular weight and size:charge ratio to be extracted. “In proteomics types of applications, that allows you to do anything from identifying amino acids and peptides to determining post-translational modifications on intact proteins, or even sequencing proteins if you can do tandem MS.”

Typically a CE experiment would involve using a silica capillary for the electrophoresis experiment and a separate pulled capillary to be used as a nanoelectrospray orifice, with the two combined through a liquid junction. The latter is what allows a potential to be applied at the distal end of the capillary to establish an electrophoretic field.

Spraying from a sharp tip allows the electric field lines to be concentrated, generating a higher electric field. Dr. Ramsey’s group has implemented the electrospray capability right onto the chip. There is a “feature in the corner of the chip where two channels meet to essentially form a liquid junction, and then a channel leading to the corner of the chip serves as an ES emitter,” he explained.

Besides pulling biomolecules along the capillaries and into an ionization chamber, electrical potential is also used to control dynamic valves—which determine whether sample flows down into the separation channel or gets shunted to a waste reservoir, replacing a more traditional mechanical mechanism.

Building on this work, Dr. Ramsey also demonstrated a dual ES emitter with two separate emitters on a single-chip structure. These can be electronically modulated on and off, allowing them to achieve the same sort of result as having two separate orifices controlled by a manual shudder—typically used to achieve higher mass accuracy.

In mass spectrometry-based proteomics, sample preparation can be critical. There is a lot of room for human error in extraction, enrichment, filtering, desalting, dialysis, and other manipulations, and discrepancies can compound and propagate throughout the process.

“If you have a lot of steps that are discontinuous and involve a lot of sample treatment in between, you’re going to lose sample and you’re going to introduce error throughout,” said Ziad El Rassi, Ph.D., chemistry professor at Oklahoma State University. “Then when you do MS comparative assessment, there is always some question whether the samples went through the same sample loss, so your comparison may not be accurate.”

Unlike DNA, there is nothing against which to “normalize” the amount of a protein in a given sample. All that can be done is compare samples to one another, and for this to have meaning samples need to be treated in the same way.

By integrating six micropreparative scale HPLC columns into a cascade-like operation via multiple switching HPLC valves, Dr. El Rassi and his colleagues are able to keep the sample moving “from one step to another without collecting, without sample treatments, without anything,” he explained. The platform allows samples to be extracted, captured, concentrated, and fractionated using a variety of techniques including affinity- and reverse-phase chromatography, all while remaining in the mobile liquid phase.

In the final chromatographic step, the sample is collected in an organic-rich mobile phase, followed by evaporating the volatile solvent, and subsequently tryptically digested. “That’s the only step required from inlet to outlet to make the sample ready for MS,” he noted.

Each run can process about 10 µL of serum, and with several runs per day, the platform allows about 100–200 µL to be processed per day, “which is a lot for LC-MS/MS,” Dr. El Rassi claimed. He demonstrated its use in the comparative analysis of the fucosylated sub-glycoproteomes of disease-free and breast cancer sera.

Yes to N- and O-

Some glycoproteins are trickier to study than others. Smaller, less complex molecules are easier to analyze by mass spectrometry, and so it would seem that it would be better to study the “glyco” part separately from the “protein” part of a glycoprotein. But unlike N-glycans (attached to asparagine), there is no enzyme that will specifically release all O-glycans (attached to serine or threonine) from the peptide backbone, so tedious chemical means need to be used, which are accompanied by complicating side reactions.

An alternative and increasingly popular approach for studying protein glycosylation is, therefore, the analysis of glycopeptides generated by proteolytic cleavage of glycoproteins.

A recent MS-based metabolomic study of urine showed that an O-glycopeptide was the discriminator between the healthy and diseased state of urinary tract infected patients. That stimulated Gerhild Zauner, Ph.D., researcher at the Leiden University Medical Center in the Netherlands, to take a closer look at the glycopeptides formed by enzymatic digestion of human fibrinogen’s backbone.

The advantage of such an approach is that the glycan portion remains attached to a peptide, yielding information on the attachment site—information that is lost when releasing the glycans. “Very often you want to have information about where the glycan is attached to the protein in order to make some sense of function or protein-protein interaction.”

Yet a specific proteinase like trypsin will often lead to long peptide stretches that may carry multiple modifications such as N- and O-glycans that can complicate analysis by MS. Therefore, “it is normally difficult to see O-glycopeptides in a trypic digest,” Dr. Zauner explained. So they opted to do a more extended study with the nonspecific Proteinase K, which can cut the backbone into much smaller pieces.

Dr. Zauner and colleagues had previously established a protocol to look at proteinase K-generated N- and O-glycopeptides, in which performing nanoLC-MS analysis using a hydrophilic interaction liquid chromatography (HILIC) column gave a clear separation of N- from O-glycopeptides. The same was found to be the case for peptides from human fibrinogen. After running the fractions through reverse-phase LC-MS, they discovered O-glycopeptides that had not been described previously.

Because Dr. Zauner didn’t know what the proteinase K fragments would be ahead of time, she “had to annotate the results by hand,” she said. As techniques to analyze O-linked glycopeptides become more widespread, instrument companies and others are developing software tools to automate the peptide and glycan assignments of glycopeptides, but for now “it’s a very time-consuming process.”

The Best of All Worlds

Even when software exists, it doesn’t always do everything a researcher needs it to.

Mass spec is often called upon to examine protein biomarkers found in biological fluids such as blood and spinal fluid, and it relies on sophisticated algorithms to do so. But “no one single method works perfectly to find the exact quantification,” said Jean Gao, Ph.D., of the University of Texas at Arlington.

The approaches currently used in proteomics apply techniques such as spectral counting, chromatographic peak area, peptide count, or sequence coverage. Yet these different methods will often produce paradoxical results.

Dr. Gao and her colleagues have been looking at the pros and cons of different approaches, and “have developed more efficient ways to integrate those different quantification methods, to obtain a more robust algorithm.” They take the measured values from each of the various methods and treat them as a feature, combining them as a vector in the new multivariate algorithm.

The data integration technique has previously been used with other technology, but not for mass spectrometry as far as Dr. Gao is aware. The advantage of high-resolution MS, she says, is that it provides more of a global view in the search for distinguishable biomarkers.

Dr. Gao and her collaborators look into subgroups of proteins that they expect to work together—those that are able to bind to albumin, for example—and compare their abundance in normal and diseased samples. The eventual clinical goal is to find how the levels of biomarkers change, she said, to discover how these proteins are contributing to the disease stages of different patients.