High-abundance protein depletion: Comparison of methods for human plasma biomarker discovery

Authors


Abstract

Affinity depletion of abundant proteins from human plasma has become a routine sample preparation strategy in proteomics used prior to protein identification and quantitation. To date, there have been limited published studies comparing the performance of commercially available depletion products. We conducted a thorough evaluation of six depletion columns using 2-DE combined with sophisticated image analysis software, examining the following criteria: (i) efficiency of high-abundance protein depletion, (ii) non-specific removal of other than the targeted proteins and (iii) total number of protein spots detected on the gels following depletion. From all the products investigated, the Seppro IgY system provided the best results. It displayed the greatest number of protein spots on the depleted plasma gels, minimal non-specific binding and high efficiency of abundant protein removal. Nevertheless, the increase in the number of detected spots compared with the second best performing and cheaper multiple affinity removal column (MARC) was not shown to be statistically significant. The ProteoPrep spin column, considered to be the “deepest” depletion technique available at the time of conducting the study, surprisingly displayed significantly fewer spots on the flow-through fraction gels compared with both the Seppro and the MARC. The spin column format and low plasma capacity were also found to be impractical for 2-DE. To conclude, we succeeded in providing an overview of the depletion columns performances with regard to the three examined areas. Our study will serve as a reference to other scientists when deciding on the optimal product for their particular projects.

1 Introduction

Human plasma is a popular biospecimen used in proteomic research as evidenced by a large number of studies being published every year. This is partly due to the fact that most, if not all, cellular proteomic changes are thought to be represented in plasma via leakage 1, 2. Thanks to additional advantageous features (accessibility, ease of processing and standardized protocols), plasma has become a common target for biomarker discovery-based projects. Nevertheless, the complexity of human plasma together with the enormous dynamic range of protein expression (>10 orders of magnitude) poses many analytical challenges 1.

There is a widespread belief that the majority of biomarkers for disease diagnosis, prognosis, monitoring of treatment and disease regression undiscovered to date are present in plasma in low concentrations as a result of dilution in blood. As a handful of proteins predominate the plasma proteomic profile and mask the signals of less abundant proteins, it appears that depletion of the major plasma proteins prior to further analysis is one of the most logical and promising preprocessing steps. It is, therefore, not surprising that various means of high-abundance protein removal have been developed and used in numerous studies as a first step in protein fractionation procedures, not limited to human plasma/serum 3–9 but also employed for cerebrospinal fluid 10–12, saliva 13 and other specimens 14–23.

From simple depletion of albumin, using either dye-ligands 24, 25 or precipitation 26–29, technology progressed towards depletion of albumin in combination with IgG by coupling bacterial proteins, Protein A/G, to a solid support 26, 27, 30. Another method employed for the removal of multiple high-abundance proteins with the aim of enriching low-molecular-weight species (peptides/proteins) was centrifugal ultrafiltration 31–34. A significant advance in the area of high-abundance protein depletion and plasma proteome characterization was achieved by applying antibodies generated against several of the most abundant plasma proteins 35–43.

Several studies comparing different depletion methods/products in terms of depletion efficiency, reproducibility, non-specific binding and enhanced detection of less abundant proteins have been undertaken and the results published 27, 28, 37, 43–52. The majority of these studies, however, concentrated on evaluating removal of albumin or albumin in combination with IgG 27, 28, 45, 50, 52. Depletion of these two most abundant plasma proteins improves the plasma protein profiling and might be sufficient in some basic studies. Nevertheless, the biomarker discovery field needs more sophisticated tools to “drill deeper” into the plasma proteome. In this respect, immunoaffinity columns designed for removal of multiple high-abundance proteins simultaneously, combined with other fractionation strategies, have seemed to be a promising solution. These antibody-based methods were usually evaluated individually by different research groups, using different methods and naturally different plasma/serum samples 35, 36, 38–41, 53. Such studies bring useful information on the performance of the various depletion columns but do not enable proper comparison. From our literature search, only a handful of comparison studies on products claiming to remove at least six high-abundance proteins have been carried out to date 37, 43, 46, 49.

In this study, we present a comprehensive evaluation of six depletion columns, including multiple affinity removal column (MARC) (human 6), Seppro and ProteoPrep depletion systems, among others. These three products represented the “deepest” depletion techniques commercially available at the time of starting the study and are still commonly used as plasma prefractionation methods. The aim of this study was to determine which of the tested depletion techniques provided the best overall performance, focusing on efficiency of high-abundance protein depletion, non-specific removal of other than the targeted proteins and enhancement in lower-abundance protein detection. Both flow-through and bound plasma fractions obtained from the affinity separation step were along with crude plasma resolved by 2-DE and gel images subjected to the analysis in the Progenesis PG 240 software. The numbers of detected protein spots were compared and found differences evaluated for significance using statistical analysis.

2 Materials and methods

2.1 Plasma sample

The human plasma sample (blood type B+) was obtained from a healthy anonymous donor (Australian Red Cross Blood Service, Sydney, NSW, Australia).

2.2 Depletion columns

Six depletion products were examined in this study, namely Aurum™ Affi-Gel® Blue mini kit (Bio-Rad, Hercules, CA, USA), Vivapure® anti-HSA/IgG kit (Sartorius Stedim Biotech, Goettingen, Germany), Qproteome albumin/IgG depletion kit (Qiagen, Hilden, Germany), MARC (human 6) (4.6×100 mm; Agilent Technologies, Santa Clara, CA, USA), Seppro® MIXED12-LC20 column (GenWay Biotech, San Diego, CA, USA) and ProteoPrep® 20 plasma immunodepletion kit (Sigma-Aldrich, St. Louis, MO, USA). Plasma proteins that are supposed to be depleted by each column, as claimed by the manufacturer, are listed in Table 1.

Table 1. List of plasma proteins that are meant to be depleted using the six evaluated depletion products
 ProteoPrepSepproMARCQproteomeVivapureAurum
Albumin
IgG 
Transferrin   
Fibrinogen    
IgA   
α-2-Macroglobulin    
IgM    
α-1-Antitrypsin   
Complement C3     
Haptoglobin   
Apolipoprotein A-I    
Apolipoprotein A-II    
Apolipoprotein B     
α-1-Acid Glycoprotein    
Ceruloplasmin     
Complement C4     
Complement C1q     
IgD     
Prealbumin     
Plasminogen     
Number of depleted proteins20126221

2.3 Depletion procedures

Depletion of high abundant plasma proteins was performed according to the manufacturer's instructions with the following changes. Proteins bound onto the Aurum column were eluted with 5 M urea, 2 M thiourea, 4% (w/v) CHAPS and 40 mM Tris. Proteins bound onto the Qproteome column were recovered with two 500 μL washes of 0.1 M glycine (pH 2.5) and neutralized with 1 M Tris (pH 9.0). Both flow-through and bound fractions obtained from the depletion procedures were stored at −80°C until processed.

2.4 Sample preparation for 2-DE

Non-depleted plasma (NDP), depleted plasma samples (flow-through fractions) as well as proteins that were bound to the affinity columns (bound fractions) were precipitated using cold acetone (−20°C) overnight (sample-to-acetone ratio, 1:4). The samples were then centrifuged at 3200g for 65 min at 10°C, and protein pellets were resuspended in 7 M urea, 2 M thiourea and 4% (w/v) CHAPS. The samples were spun again at 21 000g for 10 min at 10°C. Respective supernatants were recovered and pellets (barely visible) discarded. The samples with conductivity >300 μS/cm were further subjected to buffer exchange with 7 M urea, 2 M thiourea and 4% (w/v) CHAPS using Amicon® Ultra-4 centrifugal filter devices (Millipore, Billerica, MA, USA) with 5 kDa nominal molecular weight limit. Tris base (40 mM) was added to the samples followed by protein reduction (5 mM tributylphosphine) and alkylation (10 mM acrylamide) for 90 min at room temperature. The alkylation reaction was quenched by the addition of 10 mM DTT, and the samples were left to incubate at room temperature for 10 min. Protein concentration was then determined using FluoroProfile™ protein quantification kit (Sigma-Aldrich) with BSA used as a standard. Two separate assays were undertaken and the mean calculated.

2.5 2-DE

Three aliquots of each sample corresponding to 250 μg of proteins each were diluted with 7 M urea, 2 M thiourea and 4% (w/v) CHAPS to the volume of 317 μL. A trace (3 μL) of 0.1% (w/v) Bromophenol Blue was added into the protein solutions as a tracking dye. The samples were then loaded onto 17 cm ReadyStrip™ IPG strips (Bio-Rad) with linear pH gradient of 4–7 by passive in-gel rehydration (6 h). Isoelectric focusing was carried out on an Ettan IPGphor II (GE Healthcare) using the following program: constant step of 300 V for 4 h, gradient step from 300 to 8000 V for the next 8 h and constant step of 8000 V until a total of 110 kV h was reached. If not used immediately, the focused IPG strips were stored at −80°C.

IPG strips were equilibrated in 6 M urea, 3% (w/v) SDS, 20% (v/v) glycerol, 375 mM Tris-HCl (pH 8.8), 5 mM tributylphosphine and 10 mM acrylamide twice for 10 min on a rocker. The strips were then placed on the top of 8–18% cast gels (18 cm×18 cm) and embedded in the solution of 0.5% (w/v) agarose, 0.001% (w/v) Bromophenol Blue and SDS-PAGE running buffer (192 mM glycine, 0.1% (w/v) SDS and 24.8 mM Tris base). The second dimension separation (SDS-PAGE) was performed in a Protean II XL multi-cell (Bio-Rad) filled with the running buffer. The gels were run overnight at 4°C under the following conditions: 5 mA/gel for 4 h, 15 mA/gel for 12 h and 40 mA/gel until the Bromophenol Blue front ran off the gel.

The gels were fixed in 40% (v/v) ethanol, 10% (v/v) acetic acid and stained with Flamingo Pink (Bio-Rad) according to the manufacturer's instructions. The gels were then scanned on the Typhoon Trio Variable Mode Laser Imager (GE Healthcare) with PMT voltage set to 5 V below saturation of the most intense spot.

2.6 Image analysis

Raw gel images were uploaded into TT 900 S2S advanced alignment tool (Nonlinear Dynamics, Newcastle upon Tyne, UK) and separated into 13 groups (NDP, six flow-through fractions and six bound fractions), each consisting of three replicate gels. Based on visual examination, a gel image of the plasma sample depleted using the MARC column was chosen as a reference for the following reasons. It contained high number of spots, spots were clearly resolved without streaking and most importantly, the protein pattern appeared to have many gel areas common to both NDP and all depleted plasma gel images (with both lower and higher number of depleted proteins) which all together was assumed to simplify alignment of spots. Gel images were aligned within the replicate group and then to the reference gel. Warped images were exported to Progenesis PG 240 software (Nonlinear Dynamics) and automated spot detection was performed. Each spot was checked and manually corrected if necessary. Manual editing included merging and splitting spots, adding spots not detected by the software and deleting features (such as streaks and speckles) picked by Progenesis as spots. While editing protein spots, contrast was adjusted to the maximum so that the faintest spots were clearly visible. Further, considerable efforts were made to compare the protein patterns within the replicate groups and between the groups to ensure all the spots were included in the analysis.

2.7 Statistical analysis

The significance of differences in the numbers of protein spots detected on the NDP and flow-through fractions gel was assessed using an unpaired two-tailed Student's t-test at the significance level of α=0.05.

3 Results and discussion

The key objective of this study was to compare and evaluate the performance of commercial products designated for depletion of high-abundance plasma proteins. Combining 2-DE with sophisticated image analysis software, we assessed three criteria: (i) efficiency of high-abundance protein removal which in turn influences the degree to which the remaining proteins will be enriched and potentially detected; (ii) extent of undesirable non-specific binding which might result in partial or complete loss of proteins of interest; and (iii) total number of spots detected on the depleted plasma gels as a crude estimate of how successful the examined columns are in “uncovering” lower abundance proteins.

In our study, we did not focus on elucidating the identity of protein spots on the gels in order to assess efficiency of high-abundance protein depletion and the extent of non-specific binding. We undertook a more direct approach of matching the high-abundance proteins on our gels to those identified and made publicly available on the expasy proteomics server (http://www.expasy.ch). From the total of up to 20 high-abundance proteins claimed to be depleted using the products investigated, 12 were unambiguously localized on our gels (Fig. 1) and efficiency of their depletion evaluated. The eight remaining proteins could not be located due to their missing identifications on the Swiss-2DPAGE human plasma gel (apolipoprotein B, complement C4 (acidic) and complement C1q), their concentrations below the detection limit of 2-DE (α-1-acid glycoprotein and plasminogen), and/or the uncertainty of their position on the gels caused by the presence of other spots in the particular area (IgM, apolipoprotein A-II, IgD and complement C4 (basic)).

Figure 1.

2-DE image of NDP showing the location of 12 high-abundance proteins. The proteins were located by matching to the Swiss-2DPAGE human plasma gel (3.5–10 nonlinear IPG 18 cm, Silver stain; http://www.expasy.ch/swiss-2dpage/viewer). The position of the haptoglobin α1 chain was determined by comparison with the gels published by Mikkat et al. 60 investigating phenotype-dependent spot patterns of haptoglobin α chains.

Efficiency of high-abundance protein depletion and non-specific binding were assessed by visual examination of the flow-through (Fig. 2) and bound fractions 2-D gels (Fig. 3), respectively. Flow-through fractions gels were checked for any possible remnants of high-abundance proteins that were supposed to be depleted. Bound fractions gels were inspected for proteins depleted additionally along with the targeted high-abundance proteins. Further, numbers of spots detected on the bound fractions gels were compared and used as an indication of non-specific binding. The benefit of each depletion product on the detection of less abundant proteins was evaluated by comparing the numbers of protein spots detected on 2-D gel images before and after depletion of high-abundance proteins.

Figure 2.

2-DE images of plasma samples depleted of high abundant proteins (flow-through fractions).

Figure 3.

2-DE images of plasma proteins bound to the depletion columns (bound fractions).

3.1 Evaluation of Aurum™ Affi-Gel® Blue mini kit

The Aurum column, a representative example of Cibacron Blue-based products for albumin capture, displayed only partial depletion (Fig. 2A). Substantial amounts of albumin were not removed with corresponding spots remaining intense. Further, numerous other proteins adhered to the column (Fig. 3A). These non-specifically removed proteins corresponded to other high abundant proteins, such as IgG, fibrinogen, transferrin and apolipoprotein A-I as apparent from the comparison with the NDP gel (Fig. 1). High degree of non-specific binding was validated by the analysis in Progenesis software. An average of 581 protein spots was detected on the bound fraction gels, which represents an increase of ∼150 and ∼50 spots compared with the Vivapure column (depletion of two proteins) and the MARC column (depletion of six proteins), respectively (Table 2). The obtained results are not surprising as other published studies suggest that Cibacron Blue-based depletion of albumin is incomplete and the specificity of this ligand to albumin is poor 27, 44, 47, 51, 54, 55.

Table 2. Total number of protein spots detected on 2-D gel images
Sample1st gel2nd gel3rd gelMeanSDCV (%)
Non-depleted plasma91183488287638.894.44
Flow-through fractions
Aurum1010941101398840.734.12
Vivapure114811751145115616.521.43
Qproteome875108510511004112.7211.23
MARC127911531195120964.165.31
Seppro1154131713611277109.058.54
ProteoPrep102410061042102418.001.76
Bound fractions
Aurum62458553358145.657.86
Vivapure43052135443583.6119.22
Qproteome65556664862349.497.94
MARC50153154352521.634.12
Seppro71474658568285.2312.50
ProteoPrep78886984483441.484.98

Analysis of the flow-through fraction gels in the image analysis software resulted in the detection of 988±41 spots, which in comparison with the numbers of spots observed on the NDP gels (876±39) represents a ∼100 spots increase (Table 2). It follows that the enrichment of the remaining less abundant proteins after albumin depletion enabled the detection of new protein spots. Nevertheless, a question remains to which extent the increase in the number of spots detected was due to depletion of albumin and to which extent it was due to the considerable non-specific removal of other proteins.

3.2 Evaluation of Vivapure® anti-HSA/IgG kit

The Vivapure spin column uses an antibody fragment to bind albumin in combination with Protein G for IgG depletion. While substantial amounts of these abundant proteins were removed, a small percentage of both albumin and IgG could still be observed on the flow-through fraction gels (Fig. 2B). It is, however, important to note that the amount of the “depleted” proteins seen on the flow-through fraction gels is up to five times higher than the actual amount left in plasma after the depletion step. Proteins in both the flow-through and the bound fractions of all the examined columns were enriched so that the same amount of proteins (250 μg) could be loaded on all the gels. Theoretically, if the Vivapure column depleted albumin and IgG completely (∼80% of the total plasma protein content), this would allow for a five-fold increase in the loading of the remaining proteins.

The Vivapure column also captured several non-targeted proteins (Fig. 3B). Non-specific removal of proteins by depletion technologies in general is one of the main concerns. To a certain extent, non-specific binding is caused by strong physiological interactions between targeted and non-targeted proteins that remain unaffected under the buffer conditions used. In the study of Gundry et al. 56, authors identified 50 proteins to be associated with albumin in the retentate from the Vivapure anti-HSA column. Thirty-five of these albumin-bound proteins were further confirmed using size exclusion chromatography. In another study by Gong et al. 46, 40 proteins were found in the coimmunoprecipitate of plasma sample with an anti-HSA antibody. Finally, Zhou et al. 57 identified 210 proteins to be associated with six high-abundance proteins (albumin, IgG, IgA, IgM, apolipoprotein A-I and transferrin). Whether it is possible to further optimize buffer conditions in order to avoid non-specific removal of proteins of interest, without compromising the lifetime of antibodies, still needs to be assessed.

Depletion of one additional plasma protein (IgG) using the Vivapure kit led to further increase in the numbers of spots detected. An average of 1156 protein spots was observed on the flow-through fraction gels, with a very low CV among replicates (Table 2). CV, as a measure of 2-DE reproducibility, was also low for other samples, not exceeding 10% in the majority of cases (Table 2).

3.3 Evaluation of Qproteome albumin/IgG depletion kit

The Qproteome spin column, designed for albumin and IgG depletion through the immunoaffinity interaction with relevant antibodies, demonstrated more efficient removal of albumin compared with the Vivapure method. Only trace amounts of albumin were detected on the depleted plasma gels, enabling visualization of other comigrating proteins (Fig. 2C). Similarly to the Vivapure kit, IgG depletion was substantial although incomplete. Higher amounts of IgG seen on the Qproteome flow-through fraction gels were most likely just a direct consequence of more efficient albumin depletion allowing for an increased loading of remaining proteins, hence also IgG. Consistent with our findings, Seferovic et al. 58 had reported a complete removal of albumin and a considerable reduction of IgG levels based on Western blot results, and over 90% depletion of both albumin and IgG based on ELISA quantification.

Non-specific removal of proteins was shown to be substantially higher compared with the Vivapure column. Even though only a handful of spots corresponding to non-targeted proteins could be noticed on the Qproteome-bound fraction gels at first sight (Fig. 3C), close inspection and contrast adjustment of the gels enabled the detection of numerous additional protein spots. The total of 623±49 spots, representing an almost 200 spots difference compared with the Vivapure column (Table 2), was observed. This relatively high number of detected spots may be again attributed to more efficient depletion of albumin rather than low specificity of Qproteome antibodies. It is well known that albumin as a carrier protein interacts with many other proteins 46, 56, 57, 59. Therefore, more efficient depletion of albumin using the Qproteome column resulted in the capture of higher levels of these proteins in turn leading to downstream detection by 2-DE. Another factor contributing largely to the high values of detected spots on the bound fraction gels is the heterogeneity of immunoglobulins. The total of 238 spots identified solely as IgG was resolved on the Swiss-2DPAGE human plasma gel (3.5–10 nonlinear IPG 18 cm, Silver stain; http://www.expasy.ch/swiss-2dpage/viewer). This enormous variety of immunoglobulin sequences was most likely the major cause of the remarkably high numbers of spots detected on the bound fractions gels in general (Table 2), apart from the gels of the Aurum kit though which was the only one of the tested columns not supposed to remove IgG and other immunoglobulins.

Interestingly, the numbers of protein spots detected on the Qproteome flow-through fraction gels (1004±113) were much lower compared with the Vivapure column (1156±17). Although both of these products are designed for depletion of two plasma proteins, they use different affinity ligands, buffer conditions and depletion procedures which may be the cause of the different results obtained. Substantially higher numbers of protein spots detected on the Vivapure flow-through fraction gels also raises a question of possible protein degradation. This is very unlikely though because all the plasma samples were handled in the same way. Comparison of the Vivapure and Qproteome protein profiles, particularly concentrating on the low-molecular-weight regions of the gels (where the majority of potential protein fragments were supposed to be seen) did not reveal any major differences.

High standard deviation of the numbers of protein spots detected on triplicate Qproteome-depleted plasma gels (113 spots), corresponding to CV of 11.23% (Table 2), was due to considerably lower number of spots observed on one gel. This was most likely caused by strong background staining of this particular gel consequently hindering the detection of faint spots.

3.4 Evaluation of MARC (human 6)

The MARC column has been optimized for depletion of six high-abundance proteins via antigen–antibody interactions. The specificity of the antibodies to these six proteins appeared to be remarkably high as only a few spots present on the bound fraction gels belonged to additional proteins (Fig. 3D). Low non-specific binding of the MARC column has also been reported in other studies 47, 51, similarly using 1-D or 2-D gel electrophoresis for protein separation. Nevertheless, other reports, in which different fractionation strategies following depletion were employed (such as 2-D LC) 41, 46 or more sensitive detection methods were used (such as immunoassays) 40, provide evidence of partial or complete removal of numerous non-targeted proteins.

The removal of albumin seemed to be complete as it was not apparent on the flow-through fraction gels. This is in agreement with a study by Bjorhall et al. 55 in which at least 99.4% depletion of albumin using the MARC column was reported. The levels of IgG, transferrin, IgA and α-1-antitrypsin were greatly reduced following depletion, nevertheless all these proteins could still be detected on the flow-through fraction gels after the enrichment step. On the contrary, haptoglobin depletion was very inefficient (Fig. 2D). As previous evaluations of the MARC column regarding the efficiency of high-abundance protein depletion had been favorable 40, 44, 47, 51, 53, 55, it is possible that the leakage of the targeted proteins in the flow-through fraction was due to their relatively higher levels in this particular plasma sample. This is supported by additional 2-DE experiments that we conducted using plasma samples from three other individuals (data not shown). Two protein loadings (60 and 80 μL of plasma) were also tested. No or very low detectable quantities of all the proteins, apart from haptoglobin, were found on the depleted plasma gels. Importantly, none of the five proteins was consistently detected on all the gels suggesting that slightly different levels of these proteins in blood may affect depletion efficiency of the MARC column. Haptoglobin, on the contrary, was observed on all the flow-through fractions gels no matter the sample origin or protein amount loaded.

An average of 1209 protein spots was detected on the depleted plasma gels representing further increase compared with the other depletion products evaluated in this report earlier (Aurum, Vivapure and Qproteome columns). These data fit the observed trend that the greater the number of high-abundance proteins is removed, the greater the number of new protein spots becomes visible on the flow-through fraction 2-D gels (Table 2).

3.5 Evaluation of Seppro® MIXED12-LC20 column

The Seppro column (currently available as ProteomeLab™ IgY-12 column), using chicken IgY antibodies against 12 high-abundance proteins, displayed the greatest number of protein spots (1277±109) on the flow-through fraction gels (Fig. 2E) following depletion (Table 2). Similarly, Desrosiers et al. 49 reported higher number of spots detected on the Seppro-depleted plasma 2-D gels compared with the MARC column. In that study, spin column versions of these two columns were investigated.

The Seppro column also efficiently removed the majority of the ten claimed high-abundance proteins (IgM and α-1-acid glycoprotein were not localized on our gels as previously described) from which only minute amounts of IgG, fibrinogen, haptoglobin, α-2-macroglobulin and apolipoprotein A-I were still observable on the flow-through fraction gels. The effectiveness of depletion was directly reflected in the improved resolution of the 2-D pattern abounding in protein spots (Fig. 2E). The removal of other than the targeted proteins was negligible based on visual inspection of the respective bound fraction gels (Fig. 3E). Nevertheless, other authors 39, 46 using liquid chromatography for protein separation of depleted plasma samples reported a large number of proteins captured non-specifically along with the high-abundance proteins.

3.6 Evaluation of ProteoPrep® 20 plasma immunodepletion kit

The ProteoPrep kit from Sigma, claiming depletion of 20 high-abundance proteins, provided very positive results in terms of depletion efficiency and the extent of non-specific binding based on 2-DE results, similar to the Seppro column. From the 12 high-abundance proteins we managed to localize on our gels (Fig. 1), only traces of apolipoprotein A-I, ceruloplasmin and virtually undetectable quantities of IgG could be seen on the flow-through fraction gels (Fig. 2F).

Nevertheless, the 2-D pattern of the flow-through fraction appeared markedly less populated compared with the Seppro column, with not many new spots revealed. This was supported by the analysis of the gel images in the Progenesis software resulting in the detection of only 1024±18 protein spots. This value is much lower compared with the numbers of spots detected on the Seppro-, MARC- and even Vivapure-depleted plasma gels (Table 2). Similar results were observed by Roche et al. 43 using SELDI-TOF approach to assess the benefit of four depletion products (including MARC (human 6), ProteomeLab IgY-12 and ProteoPrep 20 spin columns) on the detection of less abundant proteins. Statistical analysis revealed significantly (P<0.05) lower number of peaks detected in the spectra of serum sample depleted using the ProteoPrep kit compared with both the MARC and the ProteomeLab columns.

We propose that considerable protein losses occurred during the depletion step as a result of increased plasma handling. With the ProteoPrep spin column plasma capacity being only 8 μL and promised ∼97% removal of the total plasma protein mass by depleting these 20 high-abundance proteins, multiple depletions had to be carried out in order to acquire sufficient quantities of the remaining less abundant proteins for running triplicate large format gels. An LC version of this column that has been commercialized recently and has higher plasma capacity (100 μL), might provide more encouraging data. It is likely that the increased plasma loading (in turn reducing the number of depletions necessary for obtaining sufficient sample) and automation would decrease observed protein losses and result in a much “richer” 2-D protein pattern.

3.7 Comparison of MARC- and Seppro-depleted plasma gels

The overlapped and aligned 2-D gel images of the plasma samples depleted using the MARC and Seppro columns, the two best performing depletion products in terms of the numbers of detected spots on the flow-through fractions gels, were further visually inspected to localize the areas of spot pattern differences (Fig. 4). Just a handful of obviously different protein spots were found on the MARC-depleted plasma gels compared with the Seppro ones, besides those corresponding to the proteins additionally removed by the Seppro column (fibrinogen, α-2-macroglobulin, IgM, apolipoprotein A-I, apolipoprotein A-II and α-1-acid glycoprotein) or those incompletely depleted by the MARC column (based on matching to the human plasma gel on the expasy server). The majority of these protein spots were observed in the low-molecular-weight region of the gels. As mentioned earlier, all the samples were treated in the same way excluding the possibility of induced degradation process. Based on the previously published reports 39, 46, 57, we suppose that these spots most likely represented small proteins non-specifically removed by the Seppro column as a result of strong interactions with the targeted proteins. Several more new spots, not seen on the gels of plasma depleted using the MARC column, were detected on the Seppro flow-through fraction gels. In this case, the new protein spots were found across the entire gel (Fig. 4), suggesting that most of these newly detected spots corresponded to lower copy number proteins which became detectable after the removal of the six additional high-abundance proteins by the Seppro column and subsequent enrichment of remaining proteins.

Figure 4.

(A) Overlaid images of the MARC (in magenta) and Seppro (in green) flow-through fractions gels (superimposition of the spots common to both gels gives rise to black color). The intense magenta spots (spots detected on the MARC-depleted plasma gel only) correspond to the high-abundance proteins additionally removed by the Seppro column and haptoglobin, IgA and transferrin incompletely depleted by the MARC column. (B) 3-D views of selected areas (A–F) showing unique spots to the Seppro-depleted plasma gel (MARC – top images; Seppro – bottom images).

3.8 Statistical analysis

In order to determine if the differences in the numbers of protein spots detected on the 2-D gels of NDP and depleted plasma samples (Table 2) were statistically significant (p<0.05), an unpaired two-tailed Student's t-test was undertaken (Table 3). The data demonstrated that the increase in the numbers of spots detected following depletion was statistically significant for all the products in the study apart from the Qproteome column. In this case, considerably low number of spots detected on one of the triplicate flow-through fraction gels contributed to the high probability p-value of the t-test.

Table 3. Student's t-test analysis of the numbers of spots detected on the non-depleted and depleted plasma gels
 NDPAurumVivapureQproteomeMARCSepproProteoPrep
  • a)

    a) Values shown correspond to the probability p of the t-test (α=0.05).

  • b)

    b) The values in bold represent statistically significant differences between the 2 means (p<0.05).

NDPx      
Aurum0.026a)b)x     
Vivapure0.00030.003x    
Qproteome0.1370.8320.081x   
MARC0.0020.0070.2380.052x  
Seppro0.0040.0130.1290.0390.403x 
ProteoPrep0.0040.2340.0010.7730.0090.017x

Further, the differences in the numbers of spots detected on the gels of plasma samples treated with individual depletion products were also compared and statistically evaluated (Table 3). Interestingly, no single column demonstrated significantly differing performance compared with all the other investigated counterparts. Indeed, grouping of a few products displaying statistically similar results could be noticed. One of the possible examples was the Vivapure, MARC and Seppro columns. Statistical analysis of the corresponding Progenesis data did not reveal any significant differences. On the contrary, the numbers of spots visible on the ProteoPrep flow-through fraction gels were proved to be significantly (p<0.05) lower compared with all these three products. The same results were obtained for the Aurum column.

Statistical analysis of the Qproteome column Progenesis data in combination with data obtained for the other tested columns did not show almost any significant differences. From all the affinity approaches examined, only Seppro depletion enabled detection of significantly (p<0.05) higher numbers of spots. Nevertheless, these findings should be looked upon with caution. As mentioned earlier, a somewhat outlier value of the number of spots detected on one of the Qproteome-depleted plasma gels might have been the cause or could have influenced the results obtained.

3.9 Cost and additional features of the examined depletion columns

The overall performance of the depletion columns should be the most important criterion when deciding on the best depletion technique. Nevertheless, there are many other additional features of the columns and factors directly associated with the depletion process, which might make some of the products preferential, depending on the particular project and its specific aims. Cost, needed instrumentation, plasma capacity, interference of depletion buffers with subsequent analysis and dilution of proteins during the depletion procedure are just a few examples of other important criteria which a potential user will also most likely take into consideration. Table 4 represents a summary of some of the above-mentioned features of the six depletion columns evaluated in our study.

Table 4. Cost and additional features of the six evaluated depletion products
 Cost per kita) (USD)Cost per 100 μL of samplea) (USD)No. of columns/kitSpin/LC columnReusabilityPlasma capacity (μL)
  • a)

    a) The values shown correspond to the cost of the products on the Australian market at the time of manuscript preparation (February 2009). Prices were converted to US dollars using the current exchange rate.

  • b)

    b) Seppro® MIXED12-LC20 column is not commercially available anymore. Instead, a smaller version of this column with 10-mL bed volume is available from Beckman Coulter (GenWay's corporate partner) sold as ProteomeLab™ IgY-12 LC10 proteome partitioning kit. Similar to the Seppro® MIXED12-LC20 column, it has plasma capacity of 250 μL and it can be reused at least 100 times. The cost (USD) of this product is $12 459 which corresponds to $49.84 per 100 μL of the sample.

Aurum™ Affi-Gel® Blue mini kit1285.1225SpinSingle use100
Vivapure® anti-HSA/IgG kit633263.7512SpinSingle use20
Qproteome albumin/IgG depletion kit268178.676SpinSingle use25
MARC – human 6, 4.6×100 mm567635.481LC≥200×60–80
Seppro® MIXED12-LC20 columnNAb)NAb)1LC>100×250
ProteoPrep® 20 plasma immunodepletion kit1129141.131Spin≥100×8

Aurum depletion, using the affinity of albumin to Cibacron Blue dye, is typically the cheapest among the products examined. Spin column format and high plasma capacity give also an advantage of many samples to be processed in a parallel fashion and within a relatively short period of time. Although the performance of the column was not shown to be outstanding, the use of this depletion technology as a fractionation technique with subsequent analysis of both flow-through and bound fractions rather than an “elimination” method may prove to be beneficial in some applications.

The Vivapure and Qproteome kits, being other examples of disposable columns, also enable quick depletion of high-abundance proteins by processing several samples at the same time. However, the plasma capacity is four to five times lower compared with the Aurum column. Therefore, multiple depletions have to be carried out in order to obtain similar amounts of proteins in the depleted fractions. Further, employment of antibodies (antibody+Protein G in case of the Vivapure column) is reflected in a relatively high cost of these products (cost of depletion of 100 μL of plasma sample was used for comparative purposes; Table 4).

The ProteoPrep column is the only example of a reusable spin column included in our study. Very low plasma capacity and no possibility for automation make this column impractical for applications where large amounts of samples are needed to be depleted, such as 2-DE. In these cases, high-abundance protein depletion using this product becomes very time-consuming and laborious. Further, many depletion runs necessary to get sufficient amount of proteins in the flow-through fraction for subsequent analysis are associated with excessive plasma sample handling. This may lead to potential protein losses as it appeared in our work and comprise reproducibility of the results.

Unlike all the other evaluated depletion products, MARC and Seppro represent LC columns. Reusability and high plasma capacity of these columns make high-abundance protein depletion relatively cheap compared with the disposable antibody-based products (Vivapure, Qproteome and ProteoPrep). Interestingly, depletion using the MARC column is the second cheapest following Aurum. Nevertheless, a very expensive LC system to which these columns have to be connected before depleting samples must be available to the user (unless a spin column version exists). Automation represents a big advantage as many samples can be depleted one after another without much manual preparation and intervention during the procedure. Samples can be easily depleted overnight. A drawback of the LC columns is substantial dilution of the samples during the depletion process. Introduction of a needed concentration step following depletion, typically performed either using a filtration device with a specific molecular weight cut-off or using solvent precipitation, may result in protein losses and affect reproducibility.

4 Concluding remarks

Our comparative study of six high-abundance protein depletion products highlights the Seppro column as a well-optimized system with the best overall performance, based on 2-DE results. The greatest number of new protein spots detected on the Seppro gels following high-abundance protein depletion makes it the most promising product for biomarker discovery studies. The MARC column which provided almost as good results as Seppro in visualizing less abundant proteins represents a more economical option and hence may become a preferred method of depletion where the budget is limited. Our work will serve as a springboard for other researchers intending to incorporate a high-abundance protein depletion method in their experimental designs.

Acknowledgements

We thank Chris Clarke and Janniche Torsvik for technical assistance with the HPLC. This research project was facilitated by access to the Australian Proteome Analysis Facility established under the Australian Government's NCRIS program.

The authors have declared no conflict of interest.