Mass spectrometric characterization of proteins transferred from polyacrylamide gels to membrane filters


H. Hirano, Yokohama City University, Advanced Medical Research Center, Fukuura 3-9, Kanazawa, Yokohama 236-0004, Japan
Fax: +81 45 787 2787
Tel: +81 45 787 2791


In the 1990s, a technique was developed to transfer proteins from electrophoresis gels onto poly(vinylidene difluoride) (PVDF) membranes, digest the proteins on the membranes with proteases such as trypsin and analyze the resulting peptides on the membranes directly by mass spectrometry to identify the proteins. This technique, based on gel electrophoresis, is particularly useful for analyzing protein isoforms, splicing variants and post-translationally modified proteins. Previously, the low ionization efficiency of peptides immobilized on the membranes often rendered this technique useless. However, this technique has been improved by the use of PVDF membranes with a small pore size, which has enabled highly efficient and effective electroblotting and mass spectrometric analyses. Here, the advantage of this technique is discussed.

Proteomic analyses begin with protein expression profiling

Proteome research involves the comprehensive analysis of protein expression profiles, the function and functional networks of proteins, and the relationship between proteins and diseases. Upgraded protein databases, and highly sensitive, accurate and high-throughput analytical techniques, such as mass spectrometry (MS), have greatly facilitated and accelerated the development of proteome research.

Two standard methods, the protein shotgun method [1] and the 2D electrophoresis (2DE)-MS/MS method [2], are commonly used in the first step of proteome analysis. In the shotgun method, a sample containing a number of proteins is digested with proteases such as trypsin, which has comparatively high substrate specificity. These digested proteins are separated by liquid chromatography (LC), analyzed by MS/MS and then identified based on their sequence information. In contrast, in the 2DE-MS/MS method, most proteins are separated by 2DE, protease-digested in gels and then analyzed by MS [3]. Next, the proteins are identified based on peptide mass fingerprints or on sequence information obtained by MS/MS and database searches [4].

Both methods can profile the expression of a number of proteins. However, the shotgun method has been more frequently used for recent proteomic analyses. The shotgun method typically identifies more proteins than 2DE-MS/MS. In addition, the shotgun method can be easily automated and therefore does not require professional skills for the analysis. In contrast, 2DE is laborious, and it is not easy for beginners to obtain reproducible results, particularly with the ‘in-gel’ digestion of proteins after electrophoresis. However, the shotgun method cannot completely replace 2DE-MS/MS for proteomic studies because 2DE-MS/MS has several advantages over the shotgun method, as described below.

Advantages of gel electrophoresis

The resolution of MS, using techniques such as matrix-assisted laser dissociation ionization-time-of-flight (MALDI-TOF)/TOF MS, electrospray ionization-quadrupole (ESI-Q)/TOF MS, ESI-linear ion trap (LIT)/TOF MS and even LTQ-Orbitrap MS, is often insufficient to measure directly the precise masses of high-molecular-weight proteins. Therefore, proteins must be digested with proteases to obtain peptides that can be analyzed using these MS approaches. Accordingly, this method often misses information on protein isoforms, splicing variants and post-translational modifications, which might be obtained by analyzing intact proteins. In contrast, electrophoresis, particularly 2DE using IEF in the first dimension and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS/PAGE) in the second dimension [5,6], can easily separate isoforms, splicing variants and post-translationally modified proteins. Jungblut et al. [7] pointed out that there are many isoforms and post-translationally modified proteins, and advocated that these protein species are a secondary component of the proteome. They noted that some proteins have tens of isoforms, including post-translationally modified proteins. These proteins can be separated by gel electrophoresis and identified by MS analysis, while it is not easy to identify these protein species using the shotgun method.

There are several useful electrophoresis techniques in addition to O’Farrell’s 2DE. For example, 2DE using Phos-tag gel electrophoresis [8] instead of SDS/PAGE in the second dimension is suitable for analyzing the phosphorylation status of phosphoproteins, which is impossible to perform using the shotgun method [9]. 2DE using blue native electrophoresis in the first dimension and SDS/PAGE in the second dimension can separate proteins within a protein complex and identify the complex components by MS [10].

In addition, a protein differential display technique using gel electrophoresis – 2D difference gel electrophoresis (2D-DIGE) – has recently been developed [11]. This technique is particularly useful for identifying disease-associated proteins [12] and post-translationally modified proteins [13].

Benefits of poly(vinylidene difluoride) membranes

Towbin et al. [14] established a western blotting technique in which proteins are separated by gel electrophoresis, electrophoretically transferred onto the appropriate membrane filters and then detected with specific antibodies. This method became widely used as a result of the high sensitivity of detection, simplicity of handling and use of straightforward techniques to desalt the samples and preserve the membranes. In addition, secondary antibodies and fluorescent reagents were recently applied to this technique, making the analysis even more sensitive and easier to perform.

Subsequently, this method was used to detect phosphoproteins [15], glycoproteins [16] and various ligand-binding proteins. At that time, nitrocellulose membranes were predominantly used for western blotting. In the 1980s, Y. Monji (Millipore, Osaka) found that poly(vinylidene difluoride) (PVDF) membranes could be used for western blotting. In those days, a novel protein sequencer, called a gas-phase sequencer, was rapidly developed that differed from previous sequencers. With the gas-phase sequencer, the purified protein was immobilized noncovalently on a small glass-fiber filter and then Edman degradation was performed on the filter. This development led to the idea that gel-resolved proteins could be transferred onto this glass-fiber filter by western blotting and sequenced using the gas-phase sequencer [17,18]. However, the blotting efficiency of the glass-fiber filter was insufficient, making this method impractical. Like the nitrocellulose membranes, the membrane filters could not be used with the sequencer because they were intolerant to the organic solvents used for Edman degradation. However, Matsudaira [19] found that PVDF membranes were suitable as a membrane filter for protein immobilization in gas-phase sequencing. Thus, PVDF membranes became widely used as a membrane filter for sequencing as well as western blotting.

Proteins blotted onto the membrane filters can be analyzed by MS

Since the 1980s, protein and peptide MS has developed rapidly. The sensitivity, accuracy and throughput have been improved by one digit per year. Presently, we can rapidly obtain information on amino acid sequences, post-translational modifications, and protein–protein interactions by MS analysis using only 1 fmol of the protein sample. In many cases, proteins are separated by 2DE, digested in gels and then identified by MS. However, this method is laborious and time consuming if an automated in-gel digestion instrument is not available. Moreover, sometimes the peptide yields are not reproducible. If proteins that are transferred from the electrophoresis gels to the membrane filters can be analyzed directly by MS, we can efficiently identify proteins on the membrane filters. For the MS analysis of proteins on membranes, it is unnecessary to perform in-gel digestion. We can analyze post-translational modifications and protein–protein interactions simultaneously by MS analyses of the proteins on the membrane filters.

Identification of proteins on membranes

Since the 1990s, studies of proteins separated by gel electrophoresis, electroblotted and analyzed by MS have been performed. In brief, proteins resolved by SDS/PAGE were electroblotted onto Fluorotrans PVDF membranes, pieces of the membrane that contained proteins were cut out and the molecular masses of the proteins on membranes were measured using MALDI-MS [20]. Subsequently, proteins separated by SDS/PAGE were electroblotted onto five types of PVDF membranes – Transblot, Immobilon PSQ, Fluorotrans, Westran S and Immobilon-P – and pieces of the membrane containing a protein of interest were incubated with the matrix, cut out and mounted on the sample support using conductive double-sided adhesive tape; then, the molecular mass values of proteins were determined by infrared (IR)-MALDI-TOF MS [21]. The authors described that in this experiment, better results were obtained from membranes that had highly specific surfaces and a low mean pore size. Similar studies were performed by several groups in order to determine the molecular mass values of gel-resolved proteins on membranes [22–24].

This technique was applied for identification of gel-resolved proteins using MALDI-MS. Patterson [25] digested gel-resolved proteins on PVDF membranes with Lys-C endopeptidase and eluted the digests from the membranes for analysis by MALDI-MS in order to identify the proteins. The author used Immobilon-CD PVDF membranes with quarternary amines linked to the membrane surface, which might be useful for efficient recovery of the electroblotted proteins (peptides) from the membranes. Later, several groups identified gel-resolved proteins using similar techniques, but different types of membranes, such as PVDF membranes (Problot and Hybond-P) and nitrocellulose [26–30].

In 1999, Schleuder et al. [31] digested gel-resolved proteins on membranes and analyzed the proteins on the membranes directly, using MALDI-MS. In brief, after digestion, they secured pieces of PVDF membrane containing the proteins onto the MALDI sample plate using conductive double-sided adhesive tape, and introduced it into the IR-MALDI-MS in order to identify the proteins by peptide mass fingerprinting. In this study, the best performance of IR-MALDI-MS was obtained using Immobilon-CD membrane, but no results were available for other PVDF membranes, such as Immobilon PSQ and Transblot membranes. In this experiment, 0.1 μm pore-size and 195-μm-thick Immobilon PSQ membranes (see Table 1 for the pore size of membranes) were used. The Immobilon-CD and 0.1 μm pore-size Immobilon PSQ membranes are not currently available commercially. Therefore, we cannot repeat these experiments.

Table 1.   Pore size of membrane filters described in this paper.
Membrane nameMembraneHydrophobicityPore size (μm)Distributor
  1. a Before 1999, 0.1 μm pore-size Immobilon PSQ membranes were available. The Immobilon PSQ membrane is 195 μm thick; the other membranes are 140–150 μm thick [21].

FluorotransPVDFHydrophobic0.2Pall (Port Washington, NY, USA)
Hybond-PPVDFHydrophobic0.45GE Healthcare (Uppsala, Sweden)
Immobilon-CDPVDFHydrophilic0.1Millipore (Bedford, MA, USA)
Immobilon FLPVDFHydrophobic0.45Millipore (Bedford, MA, USA)
Immobilon-PPVDFHydrophobic0.45Millipore (Bedford, MA, USA)
Immobilon PSQPVDFHydrophobic0.2aMillipore (Bedford, MA, USA)
0.1 μm pore-size PVDFPVDFHydrophobic0.1Millipore (Bedford, MA, USA)
ProblotPVDFHydrophobic0.2Applied Biosystems (Foster City, CA, USA)
TransblotPVDFHydrophobic0.2BioRad (Rockville, MD, USA)
Westran SPVDFHydrophobic0.2GE Healthcare (Uppsala, Sweden)

On-membrane characterization of gel-resolved proteins by MS

The gel-resolved proteins can be characterized by MALDI-MS. Sloane et al. [32] immobilized gel-resolved proteins onto Immobilon PSQ PVDF and nitrocellulose membranes, and digested the proteins with proteases on membranes using a piezoelectric chemical inkjet printer, and then they identified antigens and glycoproteins using MS. Kimura et al. [33] and Nakanishi et al. [34,35] used the same piezoelectric chemical inkjet printer for the enzymatic digestion of proteins on membranes. Kimura et al. [33] electroblotted gel-resolved proteins onto Immobilon PSQ membranes and used multiple proteases and PNGase F to identify the proteins and detect the N-linked glycans, respectively. Nakanishi et al. [35] identified tyrosine-phosphorylated proteins from Escherichia coli cells on Immobilon FL membranes using an anti-phosphotyrosine Ig. After on-membrane digestion, they identified the phosphoproteins using MALDI-TOF MS.

Chang et al. [29] separated membrane proteins from gram-negative bacteria by SDS/PAGE, electroblotted the proteins onto the Hybond-P PVDF membranes and then fixed the membrane to the sample target using conductive double-sided adhesive tape. The matrix for MALDI-TOF MS was added to the samples for MS analysis. Then, they successfully measured the molecular masses of the proteins with relative molecular mass values of 28,000-35,000. MS analysis revealed that a 28-kDa protein was N-acetylated.

Chen et al. [36] immobilized, on Hybond-P PVDF membrane, antibodies that were raised against three proteins and interacted with antigens in plasma samples. The PVDF membranes were attached to the sample plate using conductive double-sided adhesive tape and analyzed by MALDI-TOF MS. Chen et al. showed that this technique could be used to confirm the interaction between antigen and antibody, suggesting that it is possible to analyze various interactions between proteins and ligands.

Small pore size of membrane increases ion yield in MS

Thus, many interesting techniques have been developed that combine electrophoresis, blotting and MS. However, these techniques did not always yield strong and interpretable results. One of the reasons for this poor outcome is the low ionization efficiency or low ion yield of peptides and proteins that are immobilized on PVDF membranes. Therefore, it was necessary to develop new membranes on which digests of the immobilized protein can be ionized efficiently. Without this improvement, peptides obtained by on-membrane digestion of gel-resolved proteins would have to be eluted from the membrane and then subjected to MALDI for identification [25–30].

Previously, 0.2–0.45 μm pore-size PVDF membranes have been mainly used for electroblotting and subsequent MS analysis of gel-resolved proteins (Fig. 1). Recently, membranes that might be suitable for this purpose were explored [37]. In this work, proteins separated by electrophoresis were immobilized on membranes, such as PVDF, poly(tetrafluoroethylene) (PTFE), poly(ethersulfone) (PES), nitrocellulose, polycarbonate and polypropylene, by electroblotting. Proteins were not transferred efficiently onto hydrophobic membranes such as PTFE, PES, polycarbonate and polypropylene. Among these various membrane filters, PVDF membranes exhibited the highest blotting efficiency. Moreover, the efficiency of the same PVDF membrane in MS varied depending on the pore size. Among 0.1, 0.22 and 0.45 μm pore-size PVDF membranes, the mass signal intensity was highest in 0.1 μm pore-size PVDF membranes, and a smaller pore size resulted in stronger mass signal intensity was greater than that of PVDF membranes with a 0.22 μm pore-size when the proteins were digested and measured on the membranes using MALDI-TOF MS (Table 2, Fig. 2). This 0.1 μm pore-size PVDF membrane is ∼ 140 μm thick, and different from the old 195-μm-thick Immobilon PSQ membrane, as described above.

Figure 1.

 Experimental procedure of MS analysis of proteins blotted from gels onto PVDF membranes.

Table 2.   Recommended method for protein identification on membrane.
Immobilization of proteins onto membranes [39]
 Proteins were separated by electrophoresis through a polyacrylamide gel
 The polyacrylamide gel was soaked in 100 mL of 25 mm Tris/40 mm 2-amino-n-caproic acid buffer (pH 9.0) containing 20% (v/v) methanol (solution C) for 15 min with gentle shaking. This procedure was repeated twice
 PVDF membranes were dipped in 100% methanol for 10 s and subsequently in Solution C for 15 min with shaking
 200 mL of 0.3 m Tris (pH 10.5) containing 20% (v/v) methanol (Solution A), 200 mL of 25 mm Tris (pH 10.5) containing 20% (v/v) methanol (Solution B) and 200 mL of Solution C were poured into three separate stainless steel trays. Two pieces of Whatman 3MM filter paper were immersed into each solution in the trays and shaken gently for 10 min. The solution was replaced with fresh solution, and the filter papers were shaken for a further 10 min
 After removal of the excess solution attached to the filters and gel, the filter paper, PVDF membrane and gel were assembled into a sandwich as follows: from the cathode side, (a) two filter papers saturated with Solution C, (b) PVDF membranes with Solution C, (c) gel equilibrated with Solution C, (d) two filter papers saturated with Solution B and (e) two filter papers saturated with Solution A. These were sandwiched between the cathode and anode electrode plates of the semidry blotting apparatus and the proteins were transferred from the gel to the membrane
 The proteins blotted onto the PVDF membrane were detected by Direct Blue 71 staining or Ponceau-S staining
On-membrane tryptic digestion [34]
 The membranes were rinsed with distilled water and affixed to a MALDI target plate using electrically conductive double-sided adhesive tape
 An aliquot (0.5 μL) of the protease solution containing 100 μg·mL−1 of trypsin in 25 mm NH4HCO3 (pH 8.0)/40% (v/v) acetonitrile was spotted on each protein position
 Proteins on the membranes were digested with trypsin overnight at 37°C in the reaction chamber with moisture
On-membrane MS analysis [34]
 After the digestion, 0.5 μL of the matrix solution, which was saturated with α-cyano-4-hydroxycinnamic acid or 10 mg·mL−1 of 2,5-dihydroxybenzoic acid in 0.1% (v/v) trifluoroacetic acid/30% (v/v) acetonitrile, was spotted on the protein positions
 The digests were analyzed using MALDI-TOF MS or MALDI-TOF/TOF MS
Figure 2.

 Mass spectra of yeast proteins that were separated by 2DE and then electroblotted onto PVDF membranes with different pore sizes. Spots 1–4, as shown in Fig. 3, were analyzed using MALDI-quadrupole ion trap (QIT)/TOF MS.

As a trial, yeast proteins were separated by 2DE and blotted onto 0.1 μm pore-size PVDF membranes. Among hundreds of blotted proteins, 19 were selected, including high-abundance and low-abundance proteins, high-molecular-weight and low-molecular-weight proteins, and basic and acidic proteins (Fig. 3), for protein identification. After digesting these proteins with trypsin, the digests were analyzed on membranes by MALDI-TOF/TOF MS, and all 19 proteins were successfully identified. It was possible to identify comprehensively many proteins from yeast, as described previously [37].

Figure 3.

 2DE patterns of yeast proteins on gel and PVDF membrane. Yeast proteins were separated by 2DE and electroblotted onto a 0.1 μm pore-size PVDF membrane. A total of 19 distinctive proteins were digested on the membrane with trypsin. The digests were analyzed by MALDI-quadrupole ion trap (QIT)/TOF MS and the following proteins were identified: 1, phosphopyruvate hydrase; 2, fructose-bisphosphate aldolase; 3, phosphoglycerate kinase; 4, glyceraldehyde-3-phosphate dehydrogenase; 5, alcohol dehydrogenase; 6 and 7, enolase; 8, triosephosphate isomerase; 9, 5-methyltetrahydropteroyl triglutamate-homocysteine S-methyltransferase; 10, transketolase; 11, pyruvate decarboxylase; 12, inorganic pyrophosphatase; 13, thiol-specific antioxidant protein; 14, H+-transporting two-sector ATPase; 15, magnesium-activated aldehyde dehydrogenase; 16, heat shock protein SSB1; 17, DnaK-type molecular chaperone SSC1 precursor, 18; heat shock protein MSI3; and 19, heat shock protein HSP60 precursor.

It is likely that proteins cannot penetrate very small pores during electroblotting, and most of the proteins may be retained on the surface of the membrane in the case of 0.1 μm pore-size membranes. On the other hand, peptides generated by tryptic digestion may not easily spread into membranes with a small pore size. Proteins and peptides on the membrane surface are easily irradiated by lasers, resulting in increased ion yield. Furthermore, PVDF membranes with a pore size of 0.1 μm could prevent the diffusion of the matrix solution that was added to the samples.

Iwafune et al. [38] previously developed a diamond-like carbon-coated stainless steel (DLC) plate that was chemically modified with an N-hydroxysuccinimide ester. This ester reacts with primary amines on proteins and covalently attaches them to the surface of the DLC plate. Because the DLC plate is conductive, it can be used to electroblot proteins from gels following gel electrophoresis. Moreover, this plate can be used as a target plate for MALDI-TOF MS. Using the DLC plate and MALDI-TOF MS, protein digestion on the DLC plate and peptide mass fingerprinting analyses were performed to identify proteins. However, special devices are required to electroblot the proteins from the gels onto the DLC plate. For example, 0.3-mm-thick SDS-polyacrylamide gels must be used to increase the blotting efficiency, and the gels handled carefully after electrophoresis. In addition, the DLC plates are very expensive because they require complicated processing, such as surface grinding and chemical modifications, to activate the surface. Therefore, it would be ideal if membrane filters, such as PVDF, could be used instead of DLC plates.

Concluding remarks

Various biochemical characteristics of proteins can be analyzed, after separation by gel electrophoresis and electroblotting onto PVDF membranes. This technique, in which the electroblotted proteins are subjected to on-membrane protease digestion and the resultant peptides are analyzed directly by MALDI-MS on the membranes, has been studied and improved for more than 20 years and is practically useful. Developing novel blotting filters, blotting techniques and mass spectrometers will be particularly important for more efficient and effective electroblotting and MS analyses.


This work was partially supported by Special Coordination Fund for Promoting Science and Technology ‘Creation and Innovation Centers for Advanced Interdisciplinary’.