Nathan Lewis and Dr. Craig D. Thulin, Department of Chemisty and Biochemistry
Biomarkers play a key role in disease diagnosis and in guiding physicians in treatment selection. Blood serum is potentially the best tissue for biomarker discovery and detection for several reasons. First, blood is already routinely drawn for tests. Second, serum preparation is highly reproducible between differing labs because the serum preparation method is standardized. Third, the blood is in intimate contact with most other tissues and is often the route by which the body discards of cellular waste or by which endocrine signals are sent.
The low abundance, low-molecular-weight (LMW) fraction of blood serum is of particular interest in the biomarker search as it provides much information about a person’s physiological state.1,2 These species include many peptide hormones and degradative products from cellular and tissue pathophysiology. However, this fraction can be difficult to thoroughly analyze. Not only do the more highly abundant species mask these informative species, but studies have demonstrated that some of these more abundant species (such as albumin and immunoglobins) may actually bind many smaller molecules.3,4,5 Thus, it is imperative to have a handle on which portion of the serum proteome is being sacrificed in the high-abundance protein depletion method. As this is a formidable task due to the immense dynamic range of serum, this study was designed to provide an initial qualitative analysis of three methods commonly used to get at the LMW fraction.
The methods used can be found in an upcoming publication in the Journal of Proteome Research. All peaks above a S/N of three were deconvoluted to their true mass and the data sets were compared. All singly charged peaks below 1000 amu will were not analyzed at this time. These data sets were compared, and the number of distinguishable species in each sample are shown in figure 1. Each method yielded a slightly different set of masses and the percent unique to each sample is shown in Table 2. The percent uniqueness was calculated as the total number of species in the sample of question which are not detected in samples using any other method, divided by the total number of species in that sample times 100%.
As demonstrated from the results in Figure 1, each treatment caused a decrease in the total number of species detected when compared with raw serum. The raw serum samples seem to show a large number of unique species, however, many of these are likely more-abundant, less informative species. When the serum was affinity depleted with the blue sepharose and protein A/G, the remaining species are largely found also through other methods. However, the amount of unique species, 29% of the total, is large enough to warrant its use in searching for biomarkers. While the percent uniqueness of the samples depleted by the MW cutoff filters is higher than that of the affinity purified samples, the total number of unique species is less due to the small number of species identified in the size filtered samples. The most interesting find was that 65% of the species found in the samples treated with acetonitrile were unique to that method. The utility of this method is demonstrated because it has a comparable number of species found as the affinity depleted samples while yielding a larger number of unique species.
We found many of the species in the affinity purified samples also in the raw serum. The columns removed much albumin and immunoglobins (Figure 2a and 2b); however, there was still a substantial number of these species remaining as demonstrated in Figure 2b. The same result was seen with the samples purified with the spin columns (Figure 2d). When compared to the samples treated with acetonitrile, it is clear that the large highly abundant species (i.e. albumin and immunoglobins) have been more efficiently cleared and the area they normally occupy contains a number of smaller, less abundant species (Figure 2c and 2e).
If one’s goal is to find unique informative biomarkers for a particular disease, all of these methods may be employed for a robust study. The AcN precipitation shows particular promise as it effectively depletes the HMW species while concentrating down peptides and smaller molecules which may be informative degredative products or signals. If, however, it is desired to find a method for inexpensive clinical diagnosis of a disease, the AcN precipitation is quick and reproducible, provided that the biomarkers of interest are found in the AcN precipitation supernatant.
It is interesting to note that in trying to deplete the larger, more abundant species the commonly used methods to affinity depletion and spin column filtering either remove too little of the unwanted species or too much of everything else. The AcN precipitation more effectively removed the HMW species, uncovered several species normally buried in the signal of the highly abundant species, and yielded the most unique set of peaks out of the three methods. However, it is with its own challenges of possibly removing many species of interest, as well as concentrating some additional non-peptide LMW species which have posed a challenge in using this method for multivariate biomarker discovery.