Editor's note: This article was updated to clarify the work of Matthew Stone and Sricharan Bandhakavi. Their April 2010 study in which 217 novel phosphoproteins in saliva were identified increased the number of known salivary proteins, but this number was nearly doubled in an earlier study that identified more than 2,300 proteins in saliva.
Imagine this: you are sitting in your optometrist's office and he is examining your eyes. He thinks you may be colorblind, so he asks you to take a look at an Ishihara plate — a big circle filled with colored dots — and tells you that among those colored dots is a number made up of dots of a different color. If you can see it, you are not colorblind, but if you cannot, you may have a problem. So you squint your eyes, and turn your head at different angles and you do your best to pick out the number made up of green dots, when all the while, all you can see are reddish dots surrounding it.
When researchers put biological samples into a mass spectrometer and ask it to pick out the proteins it sees, that machine is very much like the patient sitting in the optometrist's chair — it sees all the red dots, but is not always able to see the green ones obscured by the red. Unfortunately for researchers, finding the green dots means more than just whether or not the mass spec is colorblind, but whether the presence of a protein it might be missing means a cancer patient likely has three months or three years to live, or if the patient will respond to monotherapy or a combination of drugs.
The rare variant, the low-abundance protein, the microsatellite — they all tell researchers something important about the disease state of a population of patients. They serve as biomarkers of disease, for diagnosis, for prognosis, for treatment outcomes, and they can also be targets for therapy. But they are all difficult to find.
In the case of low-abundance proteins, researchers are now making strides, coming up with new approaches and new technologies to strip the low-abundance proteins from biological samples in order to see exactly what they are and to figure out what they mean. Once found, these proteins can be run through the mass spec for analysis.
Of course, finding something that there isn't much of has its challenges. "The big problem in biomarkers is that the mass spectrometry itself is not very sensitive," says George Mason University's Lance Liotta, whose lab seeks to find and validate biomarkers for disease. When it comes to protein biomarkers that are only present in small amounts in various biological samples, mass spec can only do so much. "If you look at all of the different immunoassay tests that are done in a clinical lab, most of them can't be picked up by mass spec because they're just too low abundance. So even though we have a big mass spec lab, we recognize that mass spec itself — if you put the sample right in — is just not sensitive enough to discover the low abundance biomarkers," he adds.
But mass spec's vision problem is not the only hurdle. The second issue, Liotta says, is that biomarkers in serum or plasma are bound to carrier proteins like albumin, and trying to separate out and discard the carriers often means also separating out and discarding the very proteins researchers are hunting for. "You can't just put your whole serum into the mass spec, or you can't just concentrate it, because then you have just too much protein that you're overwhelming the mass spec with," Liotta says.
And there is also the issue of degradation. The proteins break down fairly quickly, so there's no time to be wasted in finding them and analyzing them, he adds.
The University of Minnesota's Matthew Stone ran into similar problems when trying to find low-abundance proteins in saliva samples. "Saliva presents issues like a lot of other biofluids where there's a large range of protein abundance," Stone says. "It's pretty easy to find proteins in saliva — the concentration is about a milligram per milliliter. So there is plenty of protein, except there are about 20 high-abundance ones that are easy to detect, and the potentially more interesting ones could be harder to detect."
The search is on
Despite the challenges, if the proteins are there, these researchers are determined to find them. In collaboration with a group of researchers in Italy, Liotta decided to take the nanotechnology route, using specialized hydrogel nanoparticles loaded with different baits to capture the proteins in one step, and in solution. This work was first published in the Journal of Materials Chemistry in August 2009. The particles are constructed of open meshwork, each of the particles is about one-fiftieth to one-one hundredth the size of a red blood cell, Liotta says. Unlike the solid nanoparticles used by many other researchers, these float in a solution and have a very fast exchange rate between what is inside them and what is all around them.
"Imagine that you have these particles floating in solution, and we can use whole blood — then, in one step, they have high affinity bait that captures the analyte of interest and traps it inside the particle and protects it from degradation because it's bound tightly," Liotta says. "And then they have a sieving surface, a specific pore size, so they exclude albumin and the immunoglobulins. In one step you have affinity, capture, cleanup, and you can pull analytes away from being bound to albumin and protect them from degradation." All a researcher has to do then is spin the tube down in a centrifuge; the nanoparticles, along with the low-abundance proteins they were built to capture, will be left behind, ready to be popped into the mass spec for analysis. This, Liotta adds, can increase the sensitivity of any measurement technology a 1,000-fold without increasing the background noise, all the while protecting the sample from degradation.
The baits that can be added to the middle of the particles are a series of organic molecules that dye the inside of the particle, and are covalently bound to the inner core. "It's like a lobster trap — it's a big, open meshwork, but then there's a trap inside that uses special dyes that bind the analyte of interest with extremely high affinity," Liotta says. "They're not specific to the analyte — they bind classes of analytes, but with high affinity and they're extremely inert, better than an antibody." In addition, these molecules pull the analyte away from albumin. One example is platelet-derived growth factor, which is usually bound to albumin in serum. However, when the serum is mixed with the nanoparticles in solution, the protein becomes attracted to the bait inside the particles, which then close around the protein, leaving the albumin out in the cold. Then, the researcher can spin down the particles or elute them.
Liotta and his group have published several papers that showcase the many ways in which this technology can be used, and how the baits may be further refined. In February, the team published a study in Biomaterials, showing the nanoparticles' use in a test to detect Lyme disease from patients' urine samples. Biomarkers of Lyme disease are notoriously difficult to find, making the illness very tricky to diagnose. But using Liotta's nanoparticle technology filled with bait that attracted the slippery antigen, the team achieved nearly 100 percent capture and 100 percent extraction of the antigen from the urine sample. The researchers are now working on creating a test that can be used in the clinic to help doctors detect Lyme disease early on.
"We have a very large number of different kinds of baits, and all different kinds of molecules — from metabolites to proteins and nucleic acids and so forth — and then the end measurement system does the precise measurement of what the analyte is," Liotta says. "We don't use antibodies routinely in the nanoparticles as the bait because oftentimes they're more expensive and not as high-affinity."
At the University of Toronto, Andrei Drabovich and Eleftherios Diamandis took a different approach to the low-abundance protein problem — which they published in the Journal of Proteome Research in March 2010 — using combinatorial peptide libraries to develop assays for monitoring low-abundance proteins. The peptide libraries help to normalize the abundance of proteins present in a sample so that low-abundance proteins are enriched and high-abundance proteins are eliminated, Drabovich says. Then, the sample can be run through mass spec for analysis.
"It works in a bit of a complicated way, but there is a notion that in this library there will be one ligand that will bind specific proteins," Drabovich says. "Let's say we have two specific proteins: a very high-abundance protein and a very low-abundance protein. When we add this combinatorial library to the sample, what happens is that there is the same number of copies of these specific ligands, so these ligands will bind to the protein. But the high-abundance proteins will over-saturate these ligands and most of them won't be bound, while low-abundance proteins will be almost completely bound to these ligands, because again, the number of ligands is equal." Then, once the sample is eluted at the end of the process, the researcher ends up with the same number of high-abundance and low-abundance proteins in the sample.
Drabovich is using this approach to search for low-abundance proteins in serum, a complex lysate of human cancer cells, and blood of ovarian and prostate cancer patients to find protein biomarkers that cannot be found by other means. "This technology is applicable to discovery and it can answer many questions, find biomarkers for diagnostic applications, for prognostic applications, for monitoring drug response," Drabovich says. "This is the first step in the whole discovery process." While the idea of the combinatorial peptide library approach is not to separate any specific protein, it will enrich those in low abundance, so that the resulting mixture can then be analyzed with mass spec technology or with other methods.
Similarly, George Mason's Liotta says there is no real need to separate specific proteins from the sample, but rather that just to separate all the low-abundance proteins is useful enough. After that, the downstream technology can take care of finding specific proteins so researchers can figure out what their presence means. "So far we haven't found the need to specifically capture one type of analyte in the particle," he says. "For human growth hormone, for example, we capture all isoforms of it at once and then elute that. Then the downstream mass spec or immunoassay or MRM can determine the specificity, so in some ways we're just increasing the concentration that goes into whatever measurement system is used."
For the University of Minnesota's Stone, saliva seemed like a good place to start to look for biomarkers for early detection of oral cancer. "Specifically, I was interested in identifying phospo-related proteins in saliva as both for basic research and cancer known to have kinase cascade activity, so some of these could be considered as biomarkers for disease," Stone says. But because saliva contains so many different proteins in high abundance, finding the rare ones required a new approach.
Stone's then-colleague at Minnesota, Sricharan Bandhakavi, suggested they use Bio-Rad's Proteominer technology to circumvent the dynamic range problem that arises when using, and thus develop saliva as a potential diagnostic fluid. "We said, 'How many proteins are there?' and 'We need to identify the low-abundance as well as the high-abundance proteins,' and 'Do they have any known roles in disease progressions?' and 'Do we have a technology that can find the low-abundance proteins so we can really study the diagnostic potential across the dynamic range?'" Bandhakavi says. In 2008, Stone, Bandhakavi, and their collaborators from four labs in the US and China, began experimenting with Proteominer to see what they could find. The result, a study published in the Journal of Proteome Research in April, describes the identification of about 217 phosphoproteins in saliva. They had previously identified about 2,340 proteins in saliva using Proteominer, nearly doubling the number of previously known saliva proteins, Bandhakavi, who now works for Bio-Rad, says. In addition, the team was able to compare the proteins in saliva to those in plasma, and found that in some diseases, protein biomarkers were more easily found in saliva samples than in plasma samples, he adds.
"[Proteominer] would be very useful for finding anything that's different from one biological state to another via differential analysis of subcellular fractions of biological fluids," Bandhakavi says. "Comparing biological states, disease states, for example cancer versus healthy, using various samples." While the biomarkers Stone and Bandhakavi found using Proteominer are more geared toward diagnosing disease in its early stages, they could eventually be used to develop targeted treatments, Bandhakavi adds.
Proteins, proteins everywhere
All of these approaches, and others being developed, can be used alone in conjunction to find as many low-abundance proteins as possible. George Mason's Liotta, his team, and his collaborators in Europe are going back over the large amounts of samples contained in a big serum biobank in Italy. Using the hydrogel nanoparticle technology, they are discovering protein biomarkers that have never been seen before. "We have a project where we're using it for saliva for head trauma," Liotta says. "[You can use it with] any body fluid, even vitreous of the eye."
Early last year, George Mason University licensed the technology to Ceres Nanosciences, and the nanoparticles loaded with various kinds of bait are being sold under the company's Nanotrap brand name, which was launched in February 2010. The particles are inert and stable, so they could also be put in common Vacutainer specimen collection tubes for blood collection, or in urine collection cups, and used by clinicians as point-of-care tests, instead of being sent to labs for analysis, Liotta says. "We have versions of it for point-of-care testing, or whenever you want to do a high-sensitivity test." There is no highly specialized equipment needed to use the nanoparticles, just some Vacutainer tubes, which most doctors use, and a centrifuge.
"The whole goal is to find the very low-abundance biomarkers and that's what we're doing," Liotta adds. "We can see things now that were completely invisible to mass spec by many, many orders of magnitude."