Validation of Free Flow Electrophoresis as a Novel Plasma and Serum Processing and Fractionation Method in Biobanking

Abstract

Free flow electrophoresis (FFE) is a fractionation method, based on isoelectric focusing (IEF). We validate the reproducibility of the method and show that it can be applied by biobanks in order to fractionate fluid biospecimens efficiently and reproducibly and to facilitate downstream proteomic applications. We also propose a simple method allowing researchers to assess the reproducibility of each FFE run.

Introduction

Despite important advances in the use of genomics and transcriptomics for the identification of clinically relevant biomarkers, circulating proteins or peptides remain the most sought-after candidate biomarkers because of ease of blood sampling and biological meaningfulness. Their initial identification in biological fluids requires proteomic analysis, usually performed on well annotated fluid samples (e.g., serum, plasma) from biobanks.

One of the most important technical difficulties in this approach is linked to the dynamic range of plasma proteins which is of the order of magnitude of 10¹⁰.¹ The most abundant protein in human plasma is serum albumin which is present at 40–50 mg/ml, while some cytokines and tissue leakage proteins are present at levels approximating 10 pg/ml. To overcome this problem, two usual strategies are followed: depletion of high abundance proteins and/or fractionation. Depletion of albumin and IgG may increase sensitivity for detection of remaining proteins, but many proteins may get irreversibly and nonspecifically bound during the process. Previous fractionation of the biological fluid is an alternative or can be combined to subsequent abundant protein depletion. One of the least explored sample fractionation methods is free flow electrophoresis (FFE).

Free flow electrophoresis, first described by Hannig,^2,3 is a prefractionation method enabling the resolution and identification of low abundant proteins and the reduction of protein mixture complexity. The method is performed with fractionation equipment that continuously streams a sample into a carrier ampholine solution flowing between two plates, as a thin laminar film (0.3–1 mm). An electrical field is applied perpendicular to the direction of flow whereby proteins and peptides are separated by IEF according to their pI values and>70 fractions are collected in a microplate. Each fraction is separated by ∼ 0.02–0.1 pH units, depending on the pH gradient created. Using this method, proteins with extreme physicochemical properties can be detected. Free flow electrophoresis has been used for pre-fractionation of samples because of two main advantages: improved sample recovery due to absence of gel media, and high sample loading capacity. Although FFE is not automatable and samples can only be processed one by one, the typical turn-around time for one sample fractionation is only 10–15 min. It has previously been shown that FFE is fit for purposes of different downstream proteomic analyses, including both targeted ELISA (enzyme-linked immunosorbent assay) and high throughput proteomic 2D-electrophoresis or mass spectrometry.^4–6 However, in order for biobanks to be able to use FFE as a routine fluid biospecimen processing/fractionation method, it is important to confirm and be able to assess the reproducibility of the method in terms of pH gradient and consequently, protein contents in each fraction or pool of fractions obtained.

Reproducibility of 2D FFE-IEF/RP-HPLC (reversed-phase high performance liquid chromatography) system has previously been studied.⁷ In that study, reproducibility was assessed from the downstream analysis point of view: RP-HPLC chromatograms, where standard deviation (SD) of retention times and of microplate well numbers were calculated across three replicates for nine peaks located between wells 13 and 87. The SD for uncorrected FFE well numbers was 0.83 (when the FFE well numbers were adjusted by pH, the SD was 0.62). This corresponds to 0.05 pH units within FFE pools. However, this study was performed with standard cellular protein preparations. The objectives of our study were to assess the reproducibility of FFE in human plasma samples, and to identify internal quality control markers that can be used to define acceptance criteria for FFE fractionation reproducibility. Therefore, we validated the reproducibility of FFE fractionation of human serum and plasma by showing that the position of proteins in a given range in a 96-well plate can be reliably predicted, enabling post-fractionation protein analysis.

Materials and Methods

Free flow electrophoresis

FFE was applied as previously described.^8–10

Preparation of the FFE buffers

Deionized and overnight degassed water was used. For the Stripe test, which ensures homogeneity of the applied electrical current, a specific dye (SPADNS; 2-(P-sulfophenylazo)-1,8-dihydroxy-3,6-naphthalene disulfate) was diluted to 0.01% in distilled and degassed water. The three separation buffers, the two stabilization buffers, and the electrode buffers were prepared according to the manufacturers' instructions for the QC2 test (BD protocol, IEF-GH 3-10, native). The 6 M urea (Serva, ref 24524) separation buffer was also prepared according to manufacturers' instructions (BD protocol, IEF 3-10, 6 M urea, denaturating conditions). The separation buffer 1 was composed of 30.3 g IEF Buffer 1, 69.7 g distilled water, and 54 g urea; the separation buffer 2 contained 100 g IEF buffer 2, 100 g distilled water, and 108 g urea; and the separation buffer 3 contained 37 g IEF buffer 3, 63 g distilled water, and 54 g urea. The three IEF buffers, the SPADNS, and the pI mix were included in the BD kit (ref. 441118). The pH and conductivities were measured after each buffer preparation. The separation buffers contained glycerol, urea, hydroxyl prolyl methyl cellulose (HPMC), H₂SO₄, Prolyte1, Prolyte 2, Prolyte 3, and NaOH. The counterflow buffer contained glycerol and urea. The anodic and cathodic circuit electrolytes were H₂SO₄ and NaOH, respectively.

Plasma and serum samples

Plasma samples and serum samples from BD Vacutainer EDTA and SST collected whole blood specimens, respectively, were prepared from two different healthy donors. The blood was centrifuged 1 h after collection at 2000 g for 10 min at room temperature, and 0.5 ml of serum or plasma was aliquoted in 1.5 ml microtubes and stored at −80°C. One aliquot for each separation was thawed on the separation day and diluted 1 to 4 in Separation buffer 2 for the IEF 3-10, 6 M urea, denaturating protocol. Final protein concentration, measured by Biorad Bradford assay, was approximately 15 mg/ml.

FFE performance test

The FFE was set up on each of five different separation days according to the manufacturer's instructions. The volume of the chamber, the media, and sample pumps flow rates were measured. A running Stripe test (QC1) and a running performance test with separation media (QC2, with pI mix) were run. The absorbance of the microplates was then measured at 405 nm to locate the SPADNS dye and the pI markers, in addition to their visual inspection. Residence time in the separation chamber was ∼ 30 min. Fractions of approximately 200 μl were collected in 96-well polypropylene plates. BD polypropylene 96-well microplates and BD polypropylene covers were used in order to avoid protein adsorption.

Validation of the method reproducibility: Five FFE separations in 6 M urea, pH 3–10

Five plasma aliquots from the same sample (plasma A) were used for the five separations to test the technical reproducibility in denaturing conditions. One aliquot of a second sample (plasma B) was used to test the biological reproducibility. The five separations were performed on 5 different days with all buffers freshly prepared. The microplates containing the FFE fractions were stored at −80°C for further analysis. Two serum aliquots were also used on 2 different days in the same separation conditions to compare plasma to serum fractionation profiles. One plasma aliquot was run in 8 M urea denaturing conditions to test for robustness of the method to composition of the separation buffer.

Electrophoretic analysis of FFE fractions

Serum and plasma FFE microplate contents were electrophoretically analyzed with the Agilent 2100 Bioanalyzer in order to detect marker peaks for assessment of FFE fractionation reproducibility. The Agilent Protein 230 Kit (in denaturing conditions) was used for the electrophoresis of the fractions according to the manufacturer's instructions. Preliminary analyses of microplate fractions showed that the most abundant protein, presumably corresponding to albumin, was detected in at least two fractions in the fraction range 4F–5G. Therefore, for each of the five separations, the microplate fractions 4F–5G were analyzed on the same chip. Collected electrophoretic data revealed six markers, each of which corresponded to a specific molecular weight and isoelectric point. For each of these protein markers, detected by Agilent electrophoresis in the fractions 4F–5G, data including the molecular weight in kDa, the relative concentrations in ng/μl, and the percentage of total protein were analyzed. The total relative concentrations of each fraction were also measured. Two different calculations were made, with or without alignment to the fraction with the highest protein concentration (presumably albumin).

Identification of internal QC protein markers

FFE separations of two plasma and two serum 96-well microplates were used for pooling and Western blotting (Wb) analysis. For each plate, three pool series were prepared: Pool 1 (wells 4F, 4G, 4H, 5A), pool 2 (wells 5B, 5C, 5D, 5E), and pool 3 (wells 5D, 5E, 5F, 5G). Each pool was made up of identical volumes of fractions from each individual well and stored at −20°C. Positive controls were prepared of plasma and serum extracted from fresh blood tubes after centrifugation at 2000 g for 10 min at room temperature and supernatant dilution with PBS 1X (1/500 and 1/100 dilution for serum and plasma respectively).

SDS-PAGE

Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) was performed using the PhastSystem^™ (GE Healthcare). Pooled samples and positive controls were diluted 1:5 in reducing SDS-PAGE sample buffer (5 ml glycerol, 0.5 ml 2-mercaptoethanol, 1 g SDS, 0.38 g Tris/HCl, 0.25 mg bromophenol Blue, 4.5 ml H₂O, pH 6.8). Samples were heated at 95°C for 2 min before electrophoresis. One μl of each pooled sample was separated on 8%–25% gradient SDS polyacrylamide gels. Molecular weight markers used were the Pierce Chemiluminescent Blue Prestained Peroxidase-labeled Protein Molecular Weight Marker. SDS-PAGE settings were 15°C gel bed temperature, 75 Vh at 250 V, 10 mA, 3 W.

Western blot analysis

Western blot analysis was performed using the BioRad Trans-Blot^® SD Semi-Dry Transfer Cell. Western Blot setup was 20 mA (constant), 20 V for 25 min. Transfer buffer contained 3.94 g Tris/HCl, 14.4 g glycine, 200 ml methanol/liter, pH 7.5. After electrotransfer, the Hybond ECL nitrocellulose membrane was incubated with ECL (chemiluminescence) Blocking Agent (GE Healthcare) for 1 h. Primary antibodies used and corresponding dilutions in Blocking Reagent were the following: anti-serum albumin, ABCAM, ab10241 (1/1000), anti-α-1-antitrypsin, ABCAM, ab49088 (1/500), anti-ceruloplasmin, ABCAM, ab48614 (1/250), anti-apolipoprotein A-I, Cell Signaling Technology, 3350S (1/500). Membranes were incubated overnight at 18°–22°C room temperature (RT), three 10 min washing steps were made in TBS1x/Tween 0.05%. Secondary antibodies used were anti-mouse IgG, horseradish peroxidase-linked, NA931 (1/5000), and anti-rabbit IgG, horseradish peroxidase-linked, NA934 (1/25000). These were diluted in TBS1x/Tween 0.05% and incubated for 1 h at RT, followed by three 10 min washing steps. Membranes were finally incubated in detection reagent (ECL plus Western blotting detection System) for 5 min and detection was performed using the ImageQuant^™ LAS 4000 (GE Healthcare).

Results

FFE performance tests

The QC1 results for each separation showed three distinct peaks of approximately the same width, 14 to 17 wells, which correspond to successful Stripe tests according to the manufacturer's guide (Figs. 1 and 2). The QC2 results also showed successful results, with seven distinct peaks corresponding to markers of specific pI (data not shown). The gradient indicated by the distribution of the pI markers of the QC2 tests, was approximately pH 5 for fractions 4F to 5G, which were analyzed with the Agilent system. The pH of all FFE buffers was stable across different preparations and separation days with separation buffers 1, 2, and 3 having pH 4, 7, and 9, respectively (Fig. 3).

FIG. 1.

(A) The FFE apparatus with horizontal separation chamber. (B) Stripe test (QC1) showing straight flow of the pH gradient.

FIG. 2.

QC1, Stripe test on six different fractionation days.

FIG. 3.

Plot showing stability of the different FFE buffer pH on different days. BD Ref corresponds to values obtained upon IQ/OQ of the equipment.

Electrophoretic profiles of the FFE fractions

Total initial plasma protein concentration was approximately 60 to 80 g/l. Comparison between total protein concentration across fractions 3A to 6G from plasma separated in 6 M urea and 8 M urea separation buffer showed similar patterns (Fig. 4). Bioanalyzer data from plasma and serum separated in 6 M urea buffer revealed six protein markers separated in three consecutive fraction pools (Fig. 5). Pool 1 (fractions 4F to 5A) showed a high percentage (28%) of a 23 kDa protein and a lower percentage (8%) of a 154 kDa protein. Mean measured protein concentrations, based on comparison of protein peak areas with kit standards, were 18.2 and 5.3 ng/μl respectively (Figs. 6A and 6F). Pool 2 (fractions 5B to 5E) analysis revealed a very high concentration (84%) of a 64 kDa protein and a very low concentration (1%) of a 119 kDa protein. The respective concentrations were 569.3 and 6.5 ng/μl (Figs. 6C and 6E). Pool 3 (fractions 5D–5G) showed peaks corresponding to a 58 kDa (15%) and a 93 kDa (12%) protein with respective concentrations of 12.5 and 16.9 ng/μl (Figs. 6B and 6D).

FIG. 4.

Plasma separation in 6 M and 8 M urea separation buffer. Total protein distribution across fractions shows the same pattern.

FIG. 5.

Validation hypothesis based on 6 protein markers separated in three different microwell pools.

FIG. 6.

Total protein concentrations (A) and distribution of each of the internal QC protein markers in microwells 4F to 5G (B, C, D, E, F, G). Average and standard deviation of concentration in ng/μl is shown for the five plasma A separations. Concentration in ng/μl is shown for plasma B separation. Well numbers were not adjusted by highest protein concentration fraction.

Therefore, six candidate internal fractionation quality control protein markers were identified at 23, 58, 64, 93, 119, 154 kD. Figure 7 shows the concentrations and percentages of total protein in each of the three consecutive fraction pools from either plasma or serum. Results are shown in microwells nonadjusted by pH (and consequently by highest protein concentration fraction), because this approach was more relevant to validation purposes. When microwells were adjusted by pH, the uncertainty was lower (results not shown). ExPaSy database searches, limited to the Homo sapiens plasma protein map, and corresponding to each molecular weight, as detected by Agilent Bioanalyzer electrophoresis, and corresponding pI ranges, as detected by the FFE separation, showed one to seven possible protein matches, corresponding to each molecular weight. For each molecular weight, we selected the protein corresponding to the highest concentration in plasma, as being the most probable match. These selected protein markers were apolipoprotein A-I (23 kDa, pI ∼ 5, expected concentration in plasma 900–2100 mg/l), α-1-antitrypsin (58 kDa, pI ∼ 4.7, expected concentration on plasma 1900–3500 mg/l), serum albumin (67 kDa, pI ∼ 5.8, expected concentration in plasma 35,000–55,000 mg/l), secretory immunoglobulin chain α (93 kDa, pI ∼ 5.3, expected concentration in plasma 900–4500 mg/l), ceruloplasmin (119 kDa and 126 kDa, pI ∼ 5.1, expected concentration in plasma 150–600 mg/l).

FIG. 7.

Concentration and total percentage of internal QC proteins in each of the three target pools, in plasma (A) and in serum (B). Upper row (23 kDa and 154 kDa proteins): target pool 4F-5A; middle row (64 kDa and 119 kDa proteins): target pool 5B-5E; low row (58 kDa and 93 kDa proteins): target pool 5D-5G.

Identification of FFE internal quality control markers

Western blot results (Fig. 8) confirmed the initial hypothesis on the identity of the potential FFE internal quality control markers except secretory immunoglobulin chain alpha. Albumin was well separated, showing bands in pool 2 and 3, in both plasma and serum samples. These pools have two fractions in common. Pool 3 contained monomeric α-1 antitrypsin. Higher molecular weight α-1 antitrypsin bands may be due to serine proteinases binding. A free α-1 antitrypsin gradient was observed comparing the three pools for both plasma and serum samples. Pool 1 contained most self-aggregated α-1 antitrypsin, whereas pool 2 and especially pool 3 contained less self-aggregated α-1 antitrypsin. Pool 3 showed very efficiently separated monomeric α-1 antitrypsin. Pool 1 contained free, monomeric apolipoprotein A-I. At low concentrations, apolipoprotein A-I can be found in a monomeric state, whereas at higher concentrations it tends to aggregate, giving rise to higher molecular weight complexes.¹¹ Pool 1 contains the most free apolipoprotein, whereas pools 2 and 3 show minor apolipoprotein A-I detection in either plasma or serum samples. Ceruloplasmin was mainly detected in two forms: abundant glycosylated (154 kDa) and low concentration nonglycosylated (119 kDa) form. It was detected in higher concentration in pool 1 (154 kDa) and lower concentration in pool 2 (119 kDa).

FIG. 8.

Representative Western blots of albumin (A), α-1-antitrypsin (B), apolipoprotein A-I (C), and ceruloplasmin (D) in plasma and serum FFE pools. 1, plasma pool of fractions 4F–5A; 2, plasma pool of fractions 5B–5E; 3, plasma pool of fractions 5D–5G; 4, plasma positive control; 5, serum pool of fractions 4F–5A; 6, serum pool of fractions 5B–5E; 7, serum pool of fractions 5D–5G; 8, serum positive control.

Based on the above results, a quality control assay for serum/plasma FFE fractionation reproducibility can thus be based on Western blotting detection of apolipoprotein A-I and/or ceruloplasmin in microwell pool 4F–5A, albumin in microwell pool 5B–5E, and α-1 antitrypsin in pool 5D–5G.

Discussion

FFE is matrix-free electrophoresis; the molecules are separated according to their charge and electrophoretic mobility in a continuous laminar flow of buffer solutions in an electric field. The laminar flow is collected in 96 capillaries at the end of the separation chamber and allows continuous fractionation in collection plates. Microwell fractions were analyzed based on our hypothesis that the fractions 4F–5G can be grouped into three pools, each containing two of the six reproducible markers indicated by Agilent Bioanalyzer (Fig. 5). Western Blot results confirmed our hypothesis.

Variation in the molecular weights of the markers may be attributed to different post-translational modifications and/or different pre-processed status of the same protein.

All of the proposed internal quality control markers were reproducibly detected in their corresponding fraction pools, with the exception of the 58 and 93 kDa markers which were absent in the 20090805 microplate. This might be due to a relatively higher flow rate of the media pump on the day of this fractionation (Fig. 9).

FIG. 9.

Effective flow rates of sample pump (μl/h, right ordinate), media pump (ml/h), and separation chamber volume (ml, left ordinate), measured on every one of the 5 plasma fractionation days.

Whereas one single ∼ 60 kDa form was detected for albumin, and found predominantly in pool 2, different molecular weight (MW) forms than the expected ones were observed for the other proteins. Pre-processed protein forms may explain these different MW forms observed on Western blots.

Apolipoprotein A-I may aggregate with other lipid components of HDL.^11,12 Pool 1 contained the expected, lower MW apolipoprotein A-I in higher concentrations than pool 2 and 3. This might be due to free apolipoprotein A-I being eluted into pool 1 wells, whereas lipid-aggregated apoA-I is eluted into other pools. This finding proves that plasma and serum fractionation by FFE is efficient and could, for example, be used for apolipoprotein A-I purification.

In ceruloplasmin Western blots, a band over 220 kDa was observed. This might be due to ceruloplasmin (122 kDa) binding to myeloperoxidase (MPO) (140 kDa).¹³ Furthermore, ceruloplasmin auto-aggregation has previously been observed.¹⁴ Pool 1 contained a 150 kDa and the>220 kDa forms, pool 2 contained the 150 kDa and a lower 60 kDa form, while pool 3 contained only the lower 60 kDa form. Ceruloplasmin contains a number of potential glycosylation sites and this may explain the higher molecular weight form of 150 kDa. The lower MW form may correspond to ceruloplasmin fragments.

Alpha-1-antitrypsin was expected to be found in higher concentrations in pool 3 than in pool 1 and 2, but this was not the case. Monomeric α-1-antitrypsin was detected in pool 3, but higher MW forms were also detected in pools 1 and 2. Residual binding of α-1 antitrypsin (52 kDa) to serine proteinases (66–75 kDa)¹⁵ may be at the origin of the higher α-1 antitrypsin molecular weight bands, observed in pools 1 and 2.

Free-flow electrophoresis shows powerful pre-fractionation capacities enabling fast and accurate analysis of a given protein mixture from a biological fluid specimen, compared to chromatography or isotachophoresis. We showed that the FFE fractions are obtained following a reproducible separation scheme. Fractionation reproducibility could be confirmed, not at a single fraction level, but at a four-fraction pool level. We propose a simple internal QC method to assess reproducibility of fractionation; this is Western blotting detection of apolipoprotein A-I and/or ceruloplasmin in fraction pool 4F–5A, albumin in fraction pool 5B–5E, and α-1 antitrypsin in fraction pool 5D–5G. Confirmation of the reproducibility of the fractionation pattern allows the biobank to sub-collect, store, and distribute specific FFE fractions or fraction pools.

Footnotes

Acknowledgments

We are grateful to the Conseil Régional de Picardie for supporting this work.

Disclosure Statement

No competing financial interests exist.

References

Anderson

, Polanski

, Pieper

, Gatlin

, Tirumalai

, Conrads

, Veenstra

, Adkins

, Pounds

, Fagan

, Lobley

. The human plasma proteome: A nonredundant list developed by combination of four separate sources. Mol Cell Proteomics, 2004; 3:311–326.

Hannig

. Die trägerfreie kontinuierliche Elektrophorese und ihre Anwendung. Fresenius Zeitscr Anal Chem, 1961; 181:244–274.

Hannig

, Heidrich

. Free Flow Electrophoresis, 1990. GIT Verlag GmbH: Darmstadt: Germany.

Hoffmann

, Ji

, Moritz

, Connolly

, Frecklington

, Layton

, Eddes

, Simpson

. Continuous free flow electrophoresis separation of cytosolic proteins from the human colon carcinoma cell line LIM 1215: A non two-dimensional gel electrophoresis-based proteome analysis strategy. Proteomics, 2001; 1:807–818.

Cho

, Lee

, Kim

, Park

, Kwon

, Park

, Lee

, Kang

, Kim

, Yoo

, Park

, Cho

, Kim

, Paik

. Efficient prefractionation of low-abundance proteins in human plasma and construction of a two-dimensional map. Proteomics, 2005; 5:3386–3396.

Moritz

, Clippingdale

, Kapp

, Eddes

, Ji

, Gilbert

, Connolly

, Simpson

. Application of 2-D free-flow electrophoresis/RP-HPLC for proteomic analysis of human plasma depleted of multi high-abundance proteins. Proteomics, 2005; 5:3402–3414.

Moritz

, Ji

, Schütz

, Connolly

, Kapp

, Speed

, Simpson

. A proteome strategy for fractionating proteins and peptides using continuous free-flow electrophoresis coupled off-line to reversed-phase high-performance liquid chromatography. Anal Chem, 2004; 76:4811–4824.

Weber

, Bocek

. Recent developments in preparative free flow isoelectric focusing. Electrophoresis, 1998; 19:1649–1653.

Moritz

, Simpson

. Liquid-based free-flow electrophoresis-reversed-phase HPLC: A proteomic tool. Nature Methods, 2005; 2:863–873.

10.

Wildgruber

, Yi

, Nissum

, Eckerskorn

, Weber

. Free-Flow Electrophoresis system for plasma proteomic applications. Methods Mol Biol, 2008; 424:287–300.

11.

Brouillette

, Anantharamaiah

, Engler

, Borhani

. Structural models of human apolipoprotein A-I: A critical analysis and review. Biochim Biophys Acta, 2001; 1531:4–46.

12.

Ajees

, Anantharamaiah

, Mishra

, Hussain

, Murthy

. Crystal structure of human apolipoprotein A-I: Insights into its protective effect against cardiovascular diseases. Proc Natl Acad Sci USA, 2006; 103:2126–2131.

13.

Segelmark

, Persson

, Hellmark

, Wieslander

. Binding and inhibition of myeloperoxidase (MPO): A major function of ceruloplasmin? Clin Exp Immunol, 1997; 108:167–174.

14.

Kang

. Modification and inactivation of human ceruloplasmin by oxidized DOPA. Bull Korean Chem Soc, 2004; 25:625–628.

15.

Kalsheker

. Alpha-antitrypsin: Structure, function and molecular biology of the gene. Biosci Rep, 1989; 9:2129–2138.