This Month's Feature

Separating Proteins by pI-Values - Can 2D LC Replace 2D GE?

By Tyge Greibrokk, Milaim Pepaj, Elsa Lundanes, Department of Chemistry, University of Oslo, Norway, Thomas Andersen, G&T Septech AS, Kolboton, Norway, and Katerina Novotna, Department of Analytical Chemistry, University of Pardubice, Czech Republic.

Separating proteins by their different isoelectric point (pI) values in the first dimension of gel electrophoresis (GE) values overcomes some of the limitations associated with two-dimensional GE. The requirements of liquid chromatography (LC) separation systems to separate proteins are often different from those of small molecules, as the inherent properties of proteins can result in loss by adsorption, precipitation at zero charge, or broad and asymmetric peaks in LC. This article looks at the specific separation problems of proteins; the advantages and disadvantages of gel electrophoresis compared with alternative techniques; the potential of pI separations by pH gradients rather than by isoelectric focusing; and describes a 2D LC technique that uses pI separation in the first dimension and reversed-phase separation in the second dimension. Separation Problems Related to Proteins Samples from living cells, tissue, or body fluids can contain thousands of different proteins, depending upon the sample and purification method. With expression levels differing by a factor of 106–108, obtaining a complete picture of the proteins in a particular sample is probably one of the most complicated separation problems of all. Highly selective sample preparation methods such as affinity chromatography (AC) often are required to reduce the complexity. Complexity is not the only problem, however. The solubility of macromolecules is not easy to foresee and control. Large hydrophobic membrane proteins have little solubility in aqueous solutions, while the smaller, more polar proteins dissolve easily. At the pH of the isoelectric point (pI), where the protein has no charge, the solubility usually is decreased, and this can lead to precipitation of the more abundant proteins. Another feature of macromolecules is the large inherent adsorption energies, resulting in frequent losses by adsorption on solid surfaces as well as on separation matrices such as column materials. Adsorption does not necessarily mean permanent loss, but because of the slow kinetics of macromolecules, this can easily lead to overlap between fractions in separation processes. Thus, trapping and separation of proteins by adsorption interactions should always involve as weak interactions as possible in order to counteract the problem caused by slow kinetics. The peak width of a protein peak in a chromatographic system is also much wider than the peak width of a small molecule, and this is at least partially related to the size of the C-term in the van Deemter equation. This means that if a protein is visualized by a sharp peak in LC, this is almost always guaranteed to be a result of gradient elution. Last but not least, the peak shape of a protein in a separation system also is dependent upon the degree of unfolding of the 3D structure. An intact globular protein will be expected to have both a different peak shape and different retention compared with a more or less unfolded structure or a completely denatured structure. There is also a risk of oxidizing thiol groups to disulphides in proteins containing cysteine or methionine groups to sulphoxides. Thus, in our opinion, there is no area of separation science with more challenges than the field of proteins, reminding the reader of the statement of Anderson and Anderson in 1998: "Considered objectively, there is every reason to expect that proteomics will ultimately exceed genomics in total effort, though this effort will sorely be limited by the availability of scientists able to deal with protein's nonideal properties." (1) Determining the Primary Structure of Proteins Determining the primary structure (the amino acid sequence) of a protein is now mainly performed by mass spectrometry (MS) or combinations of LC and MS. Reductions, alkylations, and other derivatizations are needed occasionally as well. After the matrix assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometers became available, there often appeared to be a belief that most structure determinations can be performed with these, which is a misunderstanding. The MALDI-TOF instruments have become very valuable in proteomics because of the open access format, but they cannot be used for sequencing. For this, another mass spectrometer, either a quadruple time-of-flight (Q-TOF), a triple quadrupole, an ion trap or a MALDI-TOF-TOF is required (2). This also means that a separation system based upon the requirement for extraction and transfer of stained spots from 2D GE sheets is inherently more complicated than direct transfer from an LC column to the ion source or to a tray of micro-vials (for MALDI-TOF). Two-dimensional Gel Electrophoresis (2D GE) In 2D GE, the first separation includes isoelectric focusing (IEF), allowing separation of intact proteins. Today, this is performed on an immobilized pH gradient (IPG) strip containing ampholytes to create pH gradients of wide or narrow ranges. After separation into zones by IEF, and treatment with a thiol reductant and sodium dodecyl sulphate (SDS), the strip is joined with the SDS slab gel. In a buffer containing high concentrations of chaotropic agents such as urea for unfolding the proteins, thiourea and other additives to prevent oxidation of thiol groups and reduce disulphides and detergents to keep the proteins in solution, electrophoresis is then performed on polyacrylamide gel, separating the protein-SDS complexes according to size only. Because proteins can be lost by adsorption to the IPG matrix, the presence of 2M thiourea and 5–7 M urea has been recommended in the first dimension, too (3). Although 2D GE is the standard reference method in proteomics for separation of proteins, there are many problems with the technique. Inherent problems are the lack of good quantification and of high reproducibility (4). There is also the solubility problem in the first dimension, particularly with hydrophobic proteins. Partial precipitation can lead to "smearing" of bands along the flow direction. Aggregation of proteins to larger clusters adds to that problem and lowers the resolution even more. Possibly the most serious problem is the lack of detection of the minor components because the staining procedures result in a limited dynamic range of detection. Thus, the number of proteins actually identified by 2D gels is very low. For three bacterial species, less than 5% of the proteins in SWISS-PROT were identified on the gels. Within the 5%, only between 5 and 15% were hydrophobic, compared to the theoretical hydrophobic contents of 16–29% (5). Also, in serum, biomarkers for diseases often are present at concentrations below 10 ng/mL, which makes the conventional methods inadequate for detection (6). In a human cell, the most abundant protein is often actin, with a concentration of 108 molecules per cell, while some cellular receptors or transcription factors are present at only 102–103 molecules per cell (7). (continued)