Phosphoproteome Sample Prep
by Jeffrey M. Perkel
The tool of the trade among proteomics researchers is mass spectrometry (MS). The workflow is relatively simple: Extract proteins from the cells or tissues of interest, digest them with trypsin, fractionate and inject the resulting fractions via MS.
But what about researchers interested only in the phosphoproteome? Phosphoproteins are a relatively small fraction of all proteins, and even among those, phosphopeptides represent a minority. It’s a question of abundance and stoichiometry: Not every peptide in a phosphoprotein is phosphorylated, and even those that are phosphorylated are rarely modified in 100% of all proteins. For the mass spectrometer tasked with separating the molecular wheat from the chaff, trying to focus only on phosphopeptides in a sea of unmodified peptides is a bit like sitting on the side of a busy highway and trying to photograph only a specific make and model among all the cars passing by at 75 miles per hour.
Essentially, explains William Old, who runs the proteomics facility at the University of Colorado, Boulder, some phosphopeptides are so rare in unfractionated protein extracts that mass spectrometers will never see them in a discovery mode because the instruments typically focus on the most abundant ions.
“It’s a needle-in-the-haystack problem,” Old says, and there are two ways to circumvent it. One is to train the mass spec essentially to ignore unphosphorylated peptides—an approach Old and his then-postdoctoral advisor, Natalie Ahn, used in a 2009 study in Molecular Cell that identified some 876 unique phosphorylation sites. [1] The alternative is to stack the deck in your favor, enriching the phosphopeptide population through fractionation.
A number of companies now offer tools to support such an approach. Some purify phosphoproteins, others phosphopeptides. Though there is some variation in the fractionating principle, all involve relatively simple column purifications. Whichever tool you use, though, the result is the same: a sample in which phosphorylated macromolecules are well represented, simplified to the point that they can be analyzed efficiently.
Phosphopeptide and phosphoprotein enrichment
Most proteomics experiments, including phosphoproteomics, involve peptides, rather than intact proteins. According to John Yates III, a mass spec expert who runs phosphoproteomics experiments at the Scripps Research Institute in La Jolla, Calif., phosphopeptide sample preparation typically involves two steps: fractionation and enrichment.
First, the protein extract is fractionated on either a strong cation exchange (SCX) column or a hydrophilic interaction liquid chromatography (HILIC) column. “You can use one or the other, but I think the trend has been towards HILIC,” Yates says.
HILIC, he explains, separates peptides based on their hydrophilicity; because phosphopeptides contain an extra charged group (the phosphate), they are more hydrophilic than non-phosphorylated peptides and thus are retained on the column longer. SCX fractionates by charge; phosphopeptides are less positively charged than other peptides and therefore emerge from the column earlier.
The peptide fractions that emerge from these columns are then enriched, most often by passing them over either an iron-metal affinity chromatography (IMAC) column (which sometimes use gallium instead of iron) or titanium dioxide (TiO2). Neither is really specific for phosphopeptides; IMAC works because phosphate “has an affinity for iron,” says Yates, while TiO2 binds negatively charged moieties, including some acidic glycopeptides.
In theory, either enrichment column can be used, but the two are not entirely equivalent. In a 2007 study comparing different enrichment strategies, Ruedi Aebersold, of the Swiss Federal Institute of Technology, and colleagues demonstrated that TiO2 columns exhibit a slight bias for singly phosphorylated peptides, whereas IMAC tends to favor multiply charged species. (The study also looked at a third enrichment strategy, using phosphoramidite chemistry to chemically couple phosphopeptides to a resin). Moreover, the phosphopeptide collections that the different enrichment methods pulled out were not identical; only about one-third of peptides identified by one were captured by another. [2]
One company that offers both IMAC and TiO2 phosphopeptide enrichment tools is Thermo Fisher Scientific. Both the Thermo Scientific Pierce Magnetic Phosphopeptide Enrichment Kit and TiO2 Phosphopeptide Enrichment and Clean-up Kit use TiO2 enrichment, whereas the Fe-NTA Phosphopeptide Enrichment Kit is IMAC-based.
According to Monica O’Hara Noonan, technical product manager for Thermo Scientific Pierce sample preparation products, internal data agree with Aebersold’s finding that there is a difference in the constellation of peptides each column can bind. As a result, unless researchers know for certain the phosphorylation state of the specific peptide they are interested in to begin with, she recommends using both matrices in serial—that is, passing the unbound (flow-through) material from an IMAC column over TiO2.
Sample enrichment also can occur prior to trypsination and fractionation over an SCX or HILIC column. This additional step further reduces sample complexity, says O’Hara Noonan, and it is one she recommends. “You will have a higher likelihood of seeing phosphopeptides, because you are enriching for them,” she says.
Phosphoprotein (as opposed to peptide) enrichment systems are available from Thermo Fisher, Santa Cruz Biotechnology, Qiagen and Life Technologies. The Pierce Phosphoprotein Enrichment Kit is based on IMAC but uses a different resin than Fe-NTA, according to O’Hara Noonan. Life Technologies’ Pro-Q Diamond Phosphoprotein Enrichment Kit, says product manager Nikita Warner, uses a 10-kDa molecular weight cutoff GE VivaSpin filtration concentrator, housing a phosphoprotein-binding polyether sulfone membrane, to concentrate the proteins of interest. Santa Cruz’s PhosphoCruz™ Protein Purification System and Qiagen’s PhosphoProtein Purification Kit are both affinity-based, though neither company would identify the precise nature of the purification matrix; Qiagen senior scientist Udo Roth says only that his company’s resin is “not an antibody.”
After samples emerge from phosphopeptide columns, they often must be desalted prior to injection onto a mass spectrometer, for instance on C18 columns. Salt, explains Roth, can “compromise” mass spec analysis. Though desalting tools can be purchased separately, some phosphoprotein and phosphopeptide enrichment kits bundle them in. Qiagen’s PhosphoProtein Purification Kit, for instance, includes a Nanosep Ultrafiltration Column.
The Pierce TiO2 Phosphopeptide Enrichment and Clean-up Kit includes a graphite desalting column. According to O’Hara Noonan, C18 desalting columns tend to favor hydrophobic peptides, but many phosphopeptides are relatively hydrophilic. She cites internal data in which, using C18, phosphopeptide yields were about 20%, compared with “over 80%” on graphite. “Using the graphite instead of C18 we have much, much better results and much higher yields of phosphopeptides,” she says.
Avoiding enrichment
Of course, if a mass spectrometer is fast enough, sensitive enough and accurate enough, researchers can avoid pre-enrichment altogether. In their 2009 study, Old and Ahn wanted to develop a label- and enrichment-free method for quantitative phosphoproteome analysis.
Their strategy relied on negative ion-mode scanning in a 4000 QTrap triple quadrupole-linear ion trap mass spectrometer from Applied Biosystems (now part of Life Technologies). (Phosphopeptides ionize better in negative ion mode than positive mode, the authors noted.) As peptides passed through the mass spec, they were collided with a gas, causing them to fragment; upon release of a 79-Dalton negative phosphate ion, the system would flip into positive-ion mode to sequence the peptide and then revert to negative mode to continue scanning. In this way, Old and colleagues identified 820 phosphopeptides containing 876 phosphorylation sites.
According to Old, at the time this work was conducted, the team’s yield was “in the same ballpark” of phosphoproteomics studies conducted using enrichment, which typically could identify perhaps 1,000 peptides. Since then, however, advances in mass spectrometry, especially with the high-speed, high-mass accuracy Orbitrap, have enabled researchers using enrichment to delve far deeper into the phosphoproteome; some studies have picked up as many as 20,000 sites. [3] The 4000 QTrap, by contrast, scans far too slowly to provide such extensive coverage, Old says.
As a result, for full-scale phosphoproteomics work, Old has migrated to separation and enrichment on TiO2 and other affinity strategies. He is developing other strategies as well, working to overcome the apparent biases of existing enrichment methods. But he hasn’t given up on his enrichment-free protocol; Old still applies it to samples of “medium complexity,” such as those produced by immunoprecipitation. Ultimately, though, Old hopes his method can drive more comprehensive analyses.
“This scanning method is extremely powerful,” he says. “If instrument manufacturers come up with a way to scan much more rapidly and combine that high-sensitivity ion trapping, then that method will possibly be competitive with enrichment.”
References:
[1] Old WM, et al., “Functional Proteomics Identifies Targets of Phosphorylation by B-Raf Signaling in Melanoma,” Mol Cell, 34:115-31, 2009.
[2] Bodenmiller B, et al., “Reproducible isolation of distinct, overlapping segments of the phosphoproteome,” Nat Meth, 4:231-7, 2007.
[3] Olsen JV, “Quantitative phosphoproteomics reveals widespread full phosphorylation site occupancy during mitosis,” Sci Signal, 3:ra3, 2010.