Discussion

Interest in developing improved methods for major protein depletion from serum, plasma, and other biological fluids has recently increased as proteomics technologies are sought that can aid the discovery of disease-related biomarkers. For serum or plasma this ideally requires the ability to routinely detect proteins present at ng/ mL - pg/mL levels in samples that contain a modest number of proteins at the 0.1401 mg/mL level. A logical solution would be to eliminate as many high-abundance proteins as possible so that lower abundance proteins could be detected. However, major protein depletion strategies have their critics, primarily due to the risk that proteins of interest could be removed along with the targeted proteins. A further complication is that major proteins such as albumin apparently function as carriers or molecular sponges that bind other potentially important proteins and peptides. For example, in one study when dye-based albumin depletion was used with very mild wash conditions designed to preserve noncovalent interactions, up to 63 other proteins were identified in the albumin fraction by LC-MS/MS [14].

Tab. 3 Next most abundant proteins after depletion11'

Serum

% Coverageb|

Plasma

% Coverage

Apolipoprotein A-1

86.1

Fibrinogen-g

72.4

Complement C3

86.1

Apolipoprotein A-1

65.5

Transthyretin

80.3

Hemopexin

59.3

Plasma retinol binding protein

76.1

Fibrinogen-ß

57.6

Apolipoprotein E

72.9

Plasminogen

55.3

Plasminogen

72.6

Albumin

53.7

Ribosomal protein C5

72.6

a-2-macroglobulin

53.1

Vitamin-D binding protein

72.0

Complement C3

52.4

Hemopexin

71.9

Complement C3

51.1

Apolipoprotein M

70.2

Prothrombin

51.1

Apolipoprotein A-IV

68.9

Ceruloplasmin

50

a-2-macroglobulin

68.3

Ig-a-I chain C

49

Prothrombin

65

Vitamin-D binding protein

48.9

Ceruloplasmin

64

Apolipoprotein A-IV

48.7

Tetranectin

63.4

Complement H

46.9

a) Major protein-depleted serum and plasma were fractionated using ZOOM-IEF and proteins were run on 1-D SDS-PAGE for analysis on LC-MS/MS.

b) Abundant proteins are ranked according to protein coverage as a crude estimate of abundance.

a) Major protein-depleted serum and plasma were fractionated using ZOOM-IEF and proteins were run on 1-D SDS-PAGE for analysis on LC-MS/MS.

b) Abundant proteins are ranked according to protein coverage as a crude estimate of abundance.

These results are consistent with the current study, since we show that there are large differences in the amount of nontargeted protein losses among types of affinity media and with different wash buffers. Specifically, among the resins analyzed in this study, the most extensive losses occurred with all dye-based affinity resins, while the MARS HPLC column with six polyclonal antibodies had the lowest level of nontargeted protein losses based on gel analysis.

Despite high nonspecific protein binding, dye-based affinity columns have several advantages relative to antibody-based methods. Due to their relatively low cost, they are disposable, which avoids any danger of cross-contamination between different samples. In addition, these resins are usually in a spin column format that does not require expensive HPLC equipment and that facilitates processing of samples in a cold room to minimize proteolysis. The spin column format has the added advantage that many samples or multiple aliquots of a single sample can be readily processed in parallel. Furthermore, the high nonspecific binding of dye-based affinity methods can be turned into an advantage by using the method as a fractionation step rather than a depletion step. That is, both bound and unbound fractions can be analyzed, which allows a more comprehensive analysis of the serum/plasma proteome.

The major advantage of the MARS antibody column is that it can efficiently deplete six high-abundance proteins including different molecular forms and many proteolytic products of these proteins with low nonspecific losses of other proteins. This demonstrates that it is feasible to efficiently deplete multiple major proteins in a single step with minimal losses of other proteins. Because about 85% of the total protein content of serum or plasma has been removed, this is clearly an advantage in terms of the volume equivalent of serum or plasma that can be introduced into diverse downstream analysis methods. However, the degree to which this increased load capacity contributes to detection of low-abundance proteins is highly dependent upon the downstream analysis method used. The major disadvantages of antibody-based depletion resins are those features inherent to working with antibodies, namely, relatively high cost and low sample capacity. Fortunately, antibodies are highly robust proteins, and based on many years of experience using mAb and polyclonal antibody affinity matrices for protein purification, it seems likely that such columns will last for many purification cycles if appropriate care is taken to minimize proteolysis and column clogging. In addition, as illustrated by the preliminary results described here with the prototype MARS spin column, the polyclonal antibody resin also works effectively in the spin column format, which allows parallel processing of multiple samples or aliquots and does not require complex instrumentation.

The efficient performance of the MARS column with minimal losses of target proteins apparently reflects an excellent match of specific antibodies and a specific, well-designed, proprietary binding buffer (MARS Buffer A). In contrast, other antibody affinity matrices that we tested, including the two antibody antialbumin/ IgG spin column (Figs. 1-3), an antihuman albumin monoclonal, and various chicken antibodies to human plasma proteins, showed much higher nonspecific binding (data not shown). However, the MARS Buffer A is not a universal immu-noaffinity binding buffer because it was not compatible with other antibody resins we tested (see above).

This study showed that effective depletion of six abundant proteins resulted in the ability to load larger equivalent amounts of serum or plasma into downstream separation modes including 2-D gels. But, even when 10 to 20 times more Top-6 depleted serum or plasma was applied to 2-D gels, only a modest number of new spots were detected and most of these spots were minor forms of major proteins. This was quite surprising because the silver stain we used should have a detection threshold of 0.5-1.0 ng or less, and the protein load in Fig. 4B was derived from about 25 mL. This suggests that proteins present in serum or plasma at about 20-40ng/mL or higher should be detectable if they are recovered in good yield as a single spot. But the lowest abundance proteins detected in this study are known to be present in serum at about 30 mg/mL (see above). The fact that all observed proteins were about 1000-fold more abundant than the theoretical detection limit is probably due primarily to the extensive heterogeneity of high- and medium-abundance proteins caused by extensive, variable PTMs, and physiological as well as artifactual proteolysis and oxidative damage. As a result, these abundant proteins obscure most of the 2-D gel image. For example, in the heavily loaded silver-stained gel shown in Fig. 4B, about 50% of the available separation area is heavily stained and any new low-abundance proteins that would appear in these areas could not be readily detected. Actually the only substantially open area in this gel is the low-

molecular weight region but serum and plasma contain very few proteins that are less than 30 kDa, so this is a minimally useful region ofthe gel. Hence, the extensive heterogeneity of high- and medium-abundance proteins severely limits the utility of 2-D gels for detection oflow-abundance proteins, which may often also be structurally heterogeneous and will be spread among many spots.

Even after depleting six abundant proteins, serum and plasma are extremely complex and still have a very wide range of protein abundances. To effectively mine the low-abundance regions of these proteomes, multiple high-resolution separation modes must be effectively integrated. One particularly promising approach is a new method that we are developing, which incorporates high-resolution protein and peptide separations into a 4- or 5-D protein profiling strategy (Fig. 5 [26]). This method uses three sequential separation modes to separate proteins: Top-6 protein depletion, microSol IEF, and 1-D SDS-PAGE. The 1-D gel lanes from each microSol IEF fraction are then cut into uniform slices and these latter two modes define a 2-D protein array where each point or pixel in the array contains a group of proteins with a range of known pis and a narrow range of molecular weights. Each of these gel slices is then digested with trypsin and analyzed by LC-MS/MS or LC/LC-MS/MS. Although the data shown above was obtained on the Thermo LCQ XP+, subsequent analyses on a higher sensitivity linear IT mass spectrometer (Thermo LTQ) showed dramatic increases in the number of proteins that can be identified without increasing total analysis time. Comprehensive analysis of all pixels from a human serum sample on the LTQ IT mass spectrometer using the protein array pixelation method resulted in identification of about 2400 proteins that passed the HUPO-defined stringency filter for SEQUESTdata (see Section 2). Most importantly, a number of proteins at the low nanogram per mL level could be identified [26]. Of course as with most complex peptide mixture analyses, the majority of identifications are based on single pep-tide hits and better informatics tools are needed to more reliably distinguish false positives from true positives. A recently published analysis of cerebrospinal fluid using the MARS antibody depletion column prior to shotgun MS analysis also found that this column was specific for the targeted proteins and their removal enhanced detection of lower abundance proteins [27].

In summary, efficient depletion of six abundant proteins from human serum or plasma enables the detection of more proteins with greater protein coverage when a multidimensional protein-peptide separation strategy is used. However, the next most abundant proteins rapidly become limiting both in terms of sample loading capacities and because most of the mass spectrometer time is spent identifying remaining high- and medium-abundance proteins. Hence, ideally a highly specific polyclonal antibody column that can deplete at least 18-22 ofthe most abundant proteins, which comprise 98-99% of total serum protein content, would be desirable. In this regard, during preparation of this manuscript, an immunoaffinity resin containing 12 polyclonal chicken antihuman antibodies, the Seppro™ Mixed 12 spin column (Genway Biotech, San Diego, CA, USA) became commercially available and may be a promising method of further simplifying serum and plasma for biomarker discovery.

This work was supported inpart by the National Institutes of Health Grants CA94360and CA77048 to D.W.S., and institutional grants to the Wistar Institute including an NCI Cancer Core Grant (CA10815), and the Commonwealth Universal Research Enhancement Program, Pennsylvania Department of Health. The authors would also like to thank Agilent Technologies and Sigma-Aldrichfor graciously supplying prototype products for b-test studies.

Was this article helpful?

0 0

Post a comment