1. Introduction
Sample preparation can be defined as the process that goes from the biological sample of interest to the final protein extract that can be handled by the desired proteomics method. This process has to achieve parallel goals:
• disruption of the initial structure of the biological material;
• prevention of any spurious degradation of the analyte (i.e., proteins);
• removal of compounds that can interfere with the further processing of the sample, for example, protein separation or chemical labelling or controlled digestion;
• keeping proteins soluble in conditions and media that are compatible with this downstream processing.
It can be easily seen that sample preparation strongly depends on the type of processing that will be used for the proteomics analysis per se. Two major subsets of sample preparation can be defined. In the first one, the protein-protein interactions must be kept as close as possible to the in vivo situation. This is of major importance when protein complexes are to be analyzed. In the second one, individual polypeptide chains are analyzed. In this case, separation of proteins into undegraded polypeptide chains must be achieved.
2. Native sample preparation
This means that the sample preparation must respect as much as possible the three-dimensional structure of the proteins, including the links that can exist between proteins. Whatever the subsequent separation method is, two main problems are encountered with these techniques. The first problem is to prevent degradation of the sample by hydrolases, especially proteases, which are very difficult to inhibit under the conditions that allow the three-dimensional structure of proteins to be kept intact. The second and most important problem is to keep the protein assemblies as close as possible from their in vivo situation while keeping conditions that are compatible with downstream processing.
This poses in turn the problem of choosing extraction and purification conditions that look as similar as possible to those prevailing in a cell, but still being practicable. It must be recalled that a cytosol is a 10% by weight solution of proteins, that is, a milieu of very high viscosity and of rather unknown chemical parameters (dielectric constant, ionic strength, etc.). It is therefore obvious that weakly buffered water solutions do not represent a good model of the intracellular medium and that problems with protein assemblies in such media are likely to occur. However, increasing the ionic strength is probably not adequate, as this is known to extract many subunits from complexes. The work or Schagger and coworkers (Schagger and Von Jagow, 1991) can provide interesting tracks to follow. In this paper, it is shown that high concentrations of a dielectric compound (e.g., aminocaproic acid) dramatically enhances the extraction of protein complexes. Thus, dielectric compounds could show some of the beneficial effects of salts (salting in effects), without showing their dissociating effects. However, aminocaproic acid has a relatively low dipolar moment, and might not be the ideal dielectric compound. Chemicals with stronger dipolar moments such as sulfobetaines (Vuillard et al., 1995a) have been shown to increase protein solubility and may prove useful for protein complexes isolation. Their efficiency as salt mimics has been shown for the purification of halophilic proteins (Vuillard et al., 1995b). Thus, dielectric compounds with varying dipolar moments or hydrophobic parts may be worth testing to enhance the stability of protein complexes during their fractionation either by electrophoresis or chromatographic techniques.
Apart from these general parameters, sample preparation is also driven by the separation method used. When chromatographic separations are used, for example, by affinity chromatography (Rigaut et al., 1999), the conditions are relatively flexible for many parameters such as ionic strength or detergent choice. This is not the case when native electrophoresis is used (Schagger and Von Jagow, 1991). In this case, the ionic strength must be kept low and an ionic charge must be given to the extracted proteins. This is achieved either by the use of a specially designed detergent (Hisabori et al., 1991) or via a charged protein-binding molecule such as Coomassie Blue (Schagger and Von Jagow, 1991). Apart from its use for electrophoresis, this charge-shift process has the important benefit to increase protein solubility.
3. Denaturing sample preparation: general considerations
In this case, the constraint of keeping proteins intact is replaced by separating them as much as possible in the form of individual polypeptide chains. This means in turn breaking all bonds, keeping proteins together, and allowing binding to other compounds. Apart from disulfide bridges, most of these bonds are noncovalent interactions, for example, ionic bonds, hydrogen bonds, and hydrophobic interactions. These interactions can be broken by chaotropes, detergents (especially ionic ones), or a combination thereof, while disulfide bridges must be broken by special reagents. Here again, one of the major difficulties is to prevent spurious degradation of the proteins by hydrolases. Under denaturing conditions, only very resistant hydrolases are active, that is, mainly proteases, but this can be a very important problem, especially when complete proteins are to be analyzed downstream. The problem is further enhanced by the fact that denatured, unfolded proteins are most sensitive to proteases, even if the latter are only partly active.
Last but not least, in many samples, protein denaturation induces removal of DNA-binding proteins from DNA, and thus massive swelling of DNA in the sample, correlating with increased viscosity and additional problems, depending on the downstream methods. Other nonproteinaceous compounds, such as salts or lipids, can also be deleterious for downstream methods and must be taken into account in the sample preparation protocols.
Biological samples are also generally very complex, and no single separation technique is able to resolve the sample into individual components. This means in turn that complex separation schemes, usually multidimensional (i.e., relying on different parameters), must be used. In most of these schemes, the interface process between the various separations is robust enough, so that the constraints applying to sample preparation are only those induced by the first stage of separation.
3.1. The case of disulfide bridges
Breaking of disulfide bridges is usually achieved by adding to the solubilization medium an excess of a thiol compound, most of the time dithiothreitol (DTT). However, DTT is still not a perfect reducing agent. Some proteins are not fully reduced by DTT. In addition, DTT must be used in an important excess over the protein disulfide bridges. This is a problem when thiol derivatization must be performed, as the derivatization agent will have to be present in a further excess over DTT. In these cases, phosphines are very often an effective answer. The reaction is stoichiometric, which allows us in turn to use very low concentrations of the reducing agent (a few millimolar). The most powerful compound is tributylphosphine, which was the first phosphine used for disulfide reduction in biochemistry (Ruegg and Ruidinger, 1977). However, the reagent is volatile, toxic, has a rather unpleasant odor, and needs an organic solvent to make it water-miscible. Dimethylsulfoxide (DMSO) or dimethylformamide (DMF) are suitable carrier solvents, which enable the reduction of proteins by 2-mM tributylphosphine (Kirley, 1989). These drawbacks have disappeared with the introduction of water-soluble phosphines, for example tris (carboxyethyl) phosphine, which seem however to be less potent reducers.
4. Denaturing sample preparation rationales for selected proteomics methods
The last part of this chapter will deal with sample preparation available according to the downstream proteomics approach. For space limitation reasons, only the rationales can be detailed in this chapter, and detailed protocols can be found in the references cited herein.
4.1. Sample preparation for proteomics based on zone electrophoresis
In these proteomics methods, the separation process is split in two phases (e.g., Bell et al., 2001). The first phase is a protein separation by denaturing zone electrophoresis, that is, in the presence of denaturing detergents, most often sodium dodecyl sulfate (SDS). The second phase is carried out by chromatography on the peptides produced by digestion of the separated proteins. As mentioned above, this has no impact on the sample preparation itself, which just needs to be compatible with the initial zone electrophoresis.
This is by far the simplest case. Sample preparation is achieved by mixing the initial sample with a buffered, concentrated solution of an ionic detergent, usually containing a reducer to break disulfide bridges and sometimes an additional nonionic chaotrope such as urea. Ionic detergents are among the most powerful protein denaturing solubilizing agents. Their strong binding to proteins makes all proteins to bear an electric charge of the same type, whatever their initial charge may be. This induces in turn a strong electrostatic repulsion between protein molecules, and thus maximal solubility. The system of choice is based on SDS, as this detergent binds rather uniformly to proteins. However, SDS alone at room temperature, even at high concentrations, may not be powerful enough to denature all proteins. This is why heating of the sample in the presence of SDS is usually recommended. The additional denaturation brought by heat synergizes with SDS to produce maximal solubilization and denaturation, even of the most resistant proteases.
The use of SDS is not always without drawbacks. One of the most important is encountered when the sample is rich in DNA. A terrible viscosity results, which can hamper the electrophoresis process. Moreover, some protein classes (e.g., glycoproteins) bind SDS poorly and are thus poorly separated in the subsequent electrophoresis. In such cases, it is advisable to use cationic detergents. They are usually less potent than SDS, so that a urea-detergent mixture must be used for optimal solubilization (MacFarlane, 1989). Moreover, electrophoresis in the presence of cationic detergents must be carried out at a very acidic pH, which is not technically simple but still feasible (MacFarlane, 1989). This technique has however gained recent popularity as a double zone electrophoresis method able to separate even membrane proteins (Hartinger et al., 1996).
4.2. Sample preparation for proteomics based on two-dimensional gel electrophoresis
In this scheme, the proteins are first separated by isoelectric focusing (IEF) followed by SDS electrophoresis. The constraints made on sample preparation are thus those induced by the IEF step. One of these constraints is the impossibility to use ionic detergents at high concentrations, as they would mask the protein charge and thus dramatically alter its isoelectric point (pI). Ionic detergents can however be used at low doses to enhance initial solubilization (Wilson etal., 1977), but their amount is limited by the capacity of the IEF system (in terms of ions tolerated) and by the efficiency of the detergent exchange process that takes place during the IEF step. Another major constraint induced by IEF is the requirement for low ionic strength, induced by the high electric fields required for pushing the proteins to their isoelectric points. This means in turn that only uncharged compounds can be used to solubilize proteins, that is, neutral chaotropes and detergents. The basic solubilization solutions for IEF thus contain high concentrations of a nonionic chaotrope, historically urea but now more and more a mixture of urea and thiourea (Rabilloud et al., 1997), together with a reducing agent and a nonionic detergent. While CHAPS and Triton X-100 are the most popular detergents, it has been recently shown that other detergents can enhance the solubility of proteins and give better performances (Chevallet etal., 1998; Luche etal., 2003). In this solubilization process, detergents play a multiple role. They bind to proteins and help keep them in solution, but they also break protein-lipid interactions and promote lipid solubilization. This is a problem in lipid-rich samples, in which the amount of detergent present in the sample preparation solution can be limiting. In this case, lipid removal should be included in the sample preparation process. However, lipid removal is based on protein precipitation in solvents in which lipids are soluble. The major problem is that many proteins cannot be solubilized from the precipitate under IEF-compatible conditions (e.g., Tastet et al., 2003).
Apart from lipids, IEF is also very sensitive to many other compounds, such as salts or nucleic acids, which must be removed from many samples. Salt removal is carried out either by dialysis or by precipitation of proteins (e.g., by trichloroacetic acid (TCA) or organic solvents). The classical drawback of these approaches is loss of proteins due to their sticking to the dialysis membrane in the former method and by being soluble in the precipitation medium or irreversibly precipitated in the latter method.
Nucleic acids are present at problematic concentrations in most cell extracts, not to speak of nuclear extracts. One removal method is digestion by nucleases, initially by a mixture of RNAses and DNAses (O’Farrell, 1975). As with most of enzyme-based removal methods, the main drawbacks are linked to the parallel action of proteases (Castellanos-Serra and Paz-Lago, 2002), thereby degrading the proteins, and to the addition of extraneous proteins (the nucleases). A more versatile strategy is to use a high pH during extraction, so that most proteins are anions and are repelled from the anionic nucleic acids. To avoid overswelling of the nucleic acids, which decreases the subsequent removal by ultracentrifugation, this increase of pH can be mediated by the addition of a basic polyamine (e.g., spermine, Rabilloud etal., 1994), which will precipitate the nucleic acids. However, the most basic proteins are still cations at the pH 10 obtained in the spermine extraction method, and thus stick to nucleic acids. Extraction of these basic proteins can be obtained either with competing cations such as protamine (Sanders et al., 1980) or lecithins at acid pH (Willard et al., 1979). These methods are efficient but introduce high amounts of charged compounds, so that only low sample amounts can be loaded.
Another method is based on TCA precipitation (Damerval et al., 1986), which denatures nucleic acids and gives them a positive charge, thereby repelling the proteins. While this method is able to extract basic proteins such as ribosomal proteins (Gorg et al., 1998), some proteins are lost at the resolubilization step after precipitation.
Another major problem encountered in sample preparation for two-dimensional electrophoresis resides in proteolysis. While ionic detergents are known to efficiently inactivate proteases, urea and nonionic detergents are unable to do so, as recently shown for example on yeast extracts (Harder et al., 1999). Thorough protease inactivation is clearly not an easy task. Most efficient methods are based either on TCA precipitation (with the risk of protein losses) or on the use of ionic detergents, which is always limited. It has been recently described, however, that thiourea helps in preventing spurious proteolysis (Castellanos-Serra and Paz-Lago, 2002).
4.3. Sample preparation for peptide separation-based approaches
In these methods, the sample preparation process is designed to offer optimal digestion of the proteins into peptides. This means that proteins must be extracted from the sample and denatured to maximize exposure of the protease cleavage sites. This also means that the protease used for peptide production must be active in the separation method. In this case, the robustness of many proteases is a clear advantage. Classical extraction media usually contain either multimolar concentrations of chaotropes or detergents. In the latter case, the sample is usually solubilized and denatured in high concentrations of ionic detergents, and simple dilution is used to bring the detergent concentration down to a point compatible with other steps such as chemical labeling or proteolysis.
The choice between chaotropes and detergent is driven mainly by the constraints imposed by the peptide separation method. In the wide-scope approach based on on-line two-dimensional chromatography of complex peptide mixtures (Washburn et al., 2001), both the ion exchange and reverse-phase steps are very sensitive to detergents. It must be mentioned that these on-line two-dimensional chromato-graphic methods are one of the rare cases in which the interface between the two separation methods does not bring extra robustness, so that the sample preparation must be compatible with both chromatographic methods. The modification approach (Gevaert et al., 2003), which uses extensive reverse-phase chromatog-raphy, is also very sensitive to detergent interference. This rules out the use of detergent and favors the use of chaotropes, generally nonionic ones because of the ion exchange step. Urea is used for these methods, with possible artifacts induced by urea-driven carbamylation of the sample during the lengthy digestion process. Inclusion of thiourea and lowering of the urea concentration could decrease the incidence of carbamylation in these methods. In methods in which a detergent-resistant method is used, for example, avidin selection of biotinylated peptides (Gygi et al., 1999), extraction by SDS is clearly the method of choice, as the above-mentioned drawbacks are absent.
5. Concluding remarks
It should be obvious from the above that the sample preparation process is dependent on many parameters that are sample-dependent, such as protein concentration, protein-lipid and protein-nucleic acid ratios, and so on. This explains why sample preparation, although critical for the quality of the final proteomics results, is so ill mastered and is likely to stay so.
