586 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS Aqueous Component. The hydrophobicity (and therefore retention) of a polypeptide is heavily dependent on the ionization state of the amino-acid termini (approximate pK a values of 2.4 and 9.8) and of the ionizable side chains of internal residues (pK a values in Fig. 13.1). Thus the pH of the mobile phase can have a profound effect on polypeptide retention, and it is necessary to control mobile-phase pH by the addition of a buffer (Section 7.2). In practice, separations of proteins and peptides are most often performed at low pH, where a dilute acid can serve as buffer. Under these conditions silanol ionization is suppressed, reducing undesirable interactions with protonated solutes and resulting peak tailing. Commonly used organic acids (e.g., trifluoroacetic acid) can also ion-pair with protonated solutes, resulting in increased retention for peptides and improved peak shapes for proteins. The ionization of terminal and side-chain carboxyl groups is suppressed at low pH, which further increases retention. The most widely used acid for the RPC separation of peptides and proteins is trifluoroacetic acid (TFA), with a concentration of 0.05–0.1% (approximately 5–10 mM). The stronger acidity of TFA allows a lower mobile-phase pH (≈2), and lower concentrations are therefore required compared to formic or acetic acid. A lower TFA concentration also reduces background absorbance when UV detection is used at low wavelengths. An added benefit of TFA is its high volatility, which facilitates solvent removal from collected sample fractions in preparative applications; however, purified proteins typically retain significant amounts of bound TFA, which can be removed by dialysis or diafiltration. The higher UV absorbance of organic acids necessitates detection at higher wavelengths—with increased baseline drift and noise (all of which adversely affects sensitive detection). For the trace analysis of peptides, phosphoric acid (which is transparent at 200 nm and above) can be used in place of an organic acid, but proteins often exhibit poor peak shape with phosphoric acid. The use of TFA as buffer with mass-spectrometric detection (LC-MS) can be problematic when using electrospray ionization. In negative-ion detection, the high concentration of TFA-anion can suppress solute ionization. In positive-ion detection, TFA forms such strong ion-pairs with peptides that ejection of peptide pseudomolecular ions into the gas phase is suppressed. This problem can be alleviated by the postcolumn addition of a weaker, less volatile acid such as propionic acid [17]. This ‘‘TFA fix’’ allows TFA to be used with electrospray sources interfaced with quadrupole MS systems. A more convenient solution to this TFA problem, however, is to simply replace TFA with acetic or formic acid. Tailing peaks are sometimes observed for peptides, when separated by RPC at low pH. Peak tailing is usually associated with protonated-amine groups within the solute molecule (Section 7.3.4.2). For modern, type-B alkylsilica columns, peak tailing at low pH usually depends on the weight of injected peptide, with a resulting overloading of the column (due to the mutual repulsion of positively charged solute ions in the stationary phase; Section 15.3.2.1). Column overload (and peak tailing) in these cases usually occurs for injections of > 1 μg of peptide onto a 4.6-mm ID column (or lower weights for smaller column ID’s). It is possible to inject somewhat larger weights of a peptide by increasing the ionic strength of the mobile phase; for example, by the use of fully ionized acids such as TFA or phosphoric acid, or significantly ionized buffers such as ammonium acetate or formate at higher pH [18, 19]. 13.4 SEPARATION OF PEPTIDES AND PROTEINS 587 18 20 22 24 20 22 24 26 (min) 0.02% TFA 0.2% 0.05% 0.4% 0.1% 0.8% 6 1 + 2 4 0 1 62 0 4 6 1 2 4 0 1 2 6 0 4 1 0 + 2 6 4 1 + 6 2 4 0 Figure 13.8 Effect of TFA concentration on the RPC retention of basic peptides. Sample: synthetic peptides with varying numbers of basic amino acids in the molecule (the number of these basic groups is indicated for each peak). Conditions: 250 × 4.6-mm C 18 column; acetonitrile-water gradients with indicated amount of added TFA; 26 ◦ C; 1.0 mL/min. Adapted from [20]. TFA is capable of ion-pairing with protonated, basic peptides, as illustrated in Figure 13.8. Here a mixture of synthetic peptides that contain 0, 1, 2, 4, or 6 basic residues was separated with varying concentrations of TFA. As the TFA concentration is increased, the retention of a basic peptide increases because of ion-pairing (Section 7.4.1), and the effect is greater for peptides with a larger number of basic groups. As a result relative retention changes for a change in TFA concentration. The ability to change relative retention based on peptide charge can be useful during method development. If the separation of proteins that differ in surface modifications is desired, it may be advisable to use conditions that are nondenaturing [21]. The standard, low-pH conditions described above are then inappropriate, and mobile phases buffered near neutrality are required. Buffers based on ammonium acetate, ammo- nium bicarbonate, and triethylammonium phosphate may also prove more useful in resolving polypeptide variants with differing post-translational modifications, amino-acid substitutions, or oxidation and deamidation products [21]. Triethylamine-phosphate (TEAP) as buffer, with acetonitrile as B-solvent, was recommended initially for the RPC separation of peptides [22]—primarily for the improvement of peak shape and analyte recovery. However, the current use of 588 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS type-B columns for peptide separations has largely eliminated any need for this buffer in RPC. Organic Component. The usual B-solvents for the RPC separation of peptides and proteins are acetonitrile, methanol, propanol, and isopropanol—with acetoni- trile being the most popular. The major criteria for selecting the B-solvent include low UV absorbance and viscosity, which favor acetonitrile. Solvent selectivity, cost, toxicity, and purity are also considerations. Peptides and proteins can be detected by the UV absorbance of the peptide bond at 205 to 220 nm. Acetonitrile has good optical transparency in this region, and it is compatible with detection at 205 nm. The use of methanol or propanol necessitates detection at > 210 nm, to avoid exces- sive baseline drift in gradient elution. The UV response of polypeptides decreases at longer wavelengths, with a reduction in detection sensitivity. UV detection at 205 to 220 nm is generally employed for peptides and proteins, providing ‘‘uni- versal’’ detection for all polypeptides. Tyrosine and tryptophan residues have local absorbance maxima at 270 and 280 nm, respectively, allowing the selective detection of peptides and proteins that contain these residues. However, the absorbance of the latter polypeptides at 270 to 280 nm is much lower than at 205 nm. Acetonitrile-water mixtures exhibit lower viscosity than alcohol-water, which favors faster, higher resolution separations (Section 2.3.1). The comparative behavior of acetonitrile, methanol, and isopropanol as B-solvents is illustrated in Figure 13.9 (gradient separations of a series of synthetic peptides). The analytes in this study were octapeptides of identical structure, except for different amino acids in positions 2 and 3 of the peptide [23]. Total analysis time increases in the order isopropanol < acetonitrile < methanol, suggesting that solvent strength increases in the reverse order: methanol (weakest) < acetonitrile < isopropanol (strongest). The narrowest, most symmetrical peaks are generally observed with acetonitrile as B-solvent. Solvent selectivity is different for these three B-solvents, as illustrated for a group of seven peaks marked by a bracket and an asterisk in Figure 13.9. With isopropanol as B-solvent, these peptides are poorly resolved, but their resolution increases progressively for methanol and acetonitrile. Other changes in selectivity with a change in B-solvent can also be seen in Figure 13.9. Surfactants. Surfactants are sometimes added to the sample and/or mobile phase for the solubilization of hydrophobic, poorly soluble proteins and their improved recovery from the column [21]. For example, RPC methods for integral membrane proteins often employ ionic, zwitterionic, or nonionic surfactants as additives. Since adsorbed surfactant may be difficult to remove from the column, it is advisable to dedicate a column for the use of a given surfactant. 13.4.1.3 Temperature A change in column temperature can improve the separation of peptides or proteins by RPC. First, operation at higher temperatures reduces mobile-phase viscosity and increases solute diffusion, each of which contribute to increased column efficiency and better resolution. Second, a change in column temperature can be used to optimize separation selectivity—as in the case of other ionic samples (Section 7.3.2.2). An example is provided in Figure 13.10a–d, for the gradient separation 13.4 SEPARATION OF PEPTIDES AND PROTEINS 589 (a) (b) (c) Isopropanol Acetonitrile Methanol 0 10 20 30 40 50 60 (min) Figure 13.9 Effect of B-solvent on the separation of peptides by RPC. Sample described in text. Conditions: 250 × 4.1-mm C 8 column; linear gradients at 1%B/min; pH-2; 26 ◦ C; 1.0 mL/min. Adapted from [23]. of a tryptic digest at 30 and 50 ◦ C (only peaks 6–13 shown, out of a total of 18). Finally, an increase in temperature leads to increased protein denaturation, which generally favors narrower, more symmetrical peaks and increased recovery of the sample in RPC. On the downside, operation at high temperature can reduce column lifetime for some silica-based columns. 13.4.1.4 Gradient Elution As noted above, RPC separations of biomolecules are characterized by rapid changes in retention for small changes in %B. As a consequence gradient elution is generally required for these samples. Isocratic retention k can be related approximately to mobile-phase composition (%B) by log k = log k w − Sφ (13.1) where φ (equal 0.01 × %B) is the volume-fraction of organic solvent, k w is the (extrapolated) value of k for buffer as mobile phase (φ = 0or0%B),andS is a constant for a specific solute and experimental conditions other than %B. The varying dependence of retention on %B for small molecules, peptides, and a small protein is illustrated in Figure 13.11; note that the slopes of these plots (values of S) increase with solute molecular weight M. The relationship between S and molecular weight M can be approximated by S ≈ 0.25 M 0.5 (13.1a) 590 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS 0-60% B in 75 min 45 o C; R s = 2.1 0-40% B in 50 min 45 o C; R s = 2.1 (entire chromatogram) 81012 Time (min) 0-60% B in 25 min; 30 o C R s = 1.2 (bands 11/12) 0-60% B in 75 min; 50 o C R s = 1.2 (bands 10/11) 0-60% B in 25 min; 50 o C R s = 1.4 (bands 10/11) 0-60% B in 75 min; 30 o C R s = 1.0 (bands 11/12) 18 20 22 24 26 28 30 32 Time (min) (a) (c) (e) Time (min) 6 7 8 81012 9 10 11 12 13 14 6 7 89 10 11 12 13 14 (b) (d) (f ) 16 18 20 22 24 26 28 30 Time (min) 6 7 10 11 12 13 14 8 9 10 11 12 13 14 6 7 8 9 18 20 22 24 26 28 30 Time (min) 6 7 8 9 10 11 12 14 13 02040 Time (min) 100% B 80% 60% 40% 20% 0% 1 19 6-14 Figure 13.10 Separation of rhGH peptide digest, using different gradient times and temper- atures in order to optimize selectivity and maximize resolution. Conditions: 150 × 4.6-mm C 18 column; gradients of acetonitrile (B)–water + 0.1% TFA; 2.0 mL/min; other conditions indicated in figure. Simulations based on data of [24]. In gradient elution, retention is related to solvent strength (Section 9.2) by a relationship similar to Equation (13.1): log k ∗ = log k w − Sφ ∗ (13.2) 13.4 SEPARATION OF PEPTIDES AND PROTEINS 591 1.5 1.0 0.5 0.0 -0.5 log k 20 30 40 50 % ACN/buffer benzene (M = 78) M = 4400 M = 1400 M = 13,000 M = 9,000 Solute M benzene 78 nonapeptide 1400 ACTH-(1-26) 4400 insulin 9000 cytochrome c 13000 3 9 24 31 64 S Figure 13.11 Change in isocratic retention k with change in %B as a function of solute molec- ular weight. Adapted from [25]. where k ∗ is the median value of k during gradient elution and φ ∗ is the median value of φ (value when a band has moved halfway through the column). The value of k ∗ depends on gradient conditions k ∗ = t G F V m φS (13.3) where t G is the gradient time, F is the flow rate, φ is the change in φ during the gradient (e.g., Aφ = 0.55 for a 5-60%B gradient), and V m is the column dead-volume (= t 0 F). Equation (13.3) predicts that achieving satisfactory values of k ∗ (1 ≤ k ∗ ≤ 10) for molecules with large values of S (e.g., proteins) can be accomplished by using long gradient times with (if possible) a narrow gradient range (small value of φ, i.e., a small difference in the initial and final values of %B for the gradient). Note that a change in gradient conditions that affects the value of k ∗ can change selectivity, similar to a change in %B and k in isocratic elution. Minor changes of this kind can be seen in Figure 13.10 for a change in gradient time. Similarly a change in flow rate while maintaining the same gradient time can cause changes in k ∗ and relative retention, as seen in the more dramatic example for a tryptic-peptide sample in Figure 13.12. A portion of each chromatogram in Figure 13.12a (marked by a bracket and arrow) is expanded in Figure 13.12b. 592 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS 55a 6,6a 7 5,5a 6 6a 6b 6b 7 0.5 mL/min 1.5 mL/min (b) (a) 0 10 20 30 40 50 60 (min) 0.5 mL/min 1.5 mL/min Figure 13.12 Effect of flow rate on selectivity in RPC gradient elution. Sample: peptides from the tryptic digest of myoglobin. Conditions: 80 × 6.2-mm C 8 column (5-μm particles); 10–70% ACN-buffer gradient in 60 minutes; (a) complete chromatograms; (b) expanded por- tions of chromatograms of (a ) (indicated in [a] by brackets and arrows). Adapted from [26]. Plots of log k against %B as in Figure 13.11 adequately represent polypeptide retention data for B-solvent concentrations less than 50–80%. For > 50% B, a reversal of retention may occur occasionally (e.g., Fig. 13.13). Such behavior can be observed for both peptides and proteins, and has been interpreted as a transition from RPC for low %B to hydrophilic interaction chromatography (HILIC; Sections 8.6, 13.4.4) at high %B. That is, at sufficiently high %B retention increases with further increase in %B. This phenomenon can have two practical consequences for the chromatographer. First, the extension of a gradient to organic modifier concentrations above 60–70% may sometimes be unproductive for the complete elution of polypeptides from RPC columns. Second, attempts to strip contaminating 13.4 SEPARATION OF PEPTIDES AND PROTEINS 593 10 1 0.1 0.01 k 0 20 40 60 80 100 % ACN α-endorphin leu-enkephalin ranatensin Figure 13.13 Mixed-mode retention in the RPC separation of different peptides. Conditions: C 18 column, acetonitrile-buffer mobile phases (buffer is 20 mM ammonium acetate). Adapted from [27]. proteins bound to a RPC column by the use of a high %B mobile phase may not work—which is not to say that a gradient to 100% B will often be unsuccessful. The specific behavior of Figure 13.13 is more likely to be observed for older type-A columns (Section 5.2.2.2) and positively charged solutes. Our own experience with modern type-B RPC columns suggests that increased retention at higher % B is generally unlikely. 13.4.1.5 Effect of Polypeptide Conformation The retention time of a small peptide can be estimated from its amino acid composition (Section 2.7.7 and Eq. 2.33). As polypeptide length increases beyond about 50 residues, however, such predictions become increasingly unreliable [23, 28, 29], suggesting that polypeptide conformation (which becomes more important for larger molecules) can play an important role in RPC retention and separation. For peptides that contain proline, the slow interconversion of cis and trans configurations of this amino acid can give rise to peak broadening [30] or even complete resolution of the two conformers [31]. The combination of low pH, the presence of organic solvent in the mobile phase, and hydrophobic RPC columns creates a denaturing environment for polypeptides, which is further enhanced at higher temperatures. Conformational changes during migration of a protein through the column can have an adverse effect on protein separation. The injection of a protein into a hydrophobic RPC column will result in a total loss of quaternary structure, and partial or complete loss of tertiary structure. Denaturation of the protein exposes hydrophobic residues normally sequestered within the interior of the native protein, with a resulting increase in protein retention. If partial denaturation occurs prior to chromatographic migration—and if further denaturation is relatively slow—the resolution of native from fully denatured protein can result in the appearance of 594 BIOCHEMICAL AND SYNTHETIC POLYMER SEPARATIONS 20°C 25°C 30°C 37°C 10 15 10 15 10 15 10 15 (min) Figure 13.14 Effect of temperature on peak shape for ribonuclease. Conditions: 100 × 4.6-mm (10 μm) LiChrospher C4 column; solvent A, 10 mM H 3 PO 4 , pH-2.2; sol- vent B, 55/45 H 2 O/1-propanol + 10 mM H 3 PO 4 ; 5–85% B in 30 min; flow rate, 1 mL/min. Adapted from [32]. two distinct peaks. If relatively slow denaturation occurs during migration, the two forms may overlap, resulting in the appearance of a broad, misshapen peak. The slow interconversion of conformers can also cause increased peak broadening without the appearance of two peaks. Slow denaturation is associated with poor recovery of sample mass and ‘‘ghost peaks’’ (elution of the same sample in subsequent injections). Partial denaturation during RPC separation as a function of temperature [32] was first demonstrated for the small protein ribonuclease (Fig. 13.14). At low temperatures, the chromatogram exhibits a sharp, late-eluting peak with a broad, early-eluting shoulder. Spectral measurements indicate that the sharp peak is the denatured molecule. The broad shoulder represents native protein generated by refolding of ribonuclease during elution. Chromatography at elevated temperatures favors full denaturation immediately upon injection, with conversion of the protein into a single, more hydrophobic (and more retained) species. Separation problems in RPC due to conformational effects can be minimized by the use of conditions that favor the denatured state; for example, operation at elevated temperatures with a more hydrophobic column and low-pH mobile phase. The use of sample pretreatment with denaturing conditions and separation at 60 ◦ C has been used to obtain acceptable peak shapes and better recoveries for a variety of proteins differing in size and isoelectric points [33, 34]. See also the example of Figure 6.21 of [3], where recovery and peak shape for a recombinant protein was monitored from 50 to 90 ◦ C; peak shape continued to improve at higher temperatures, but recovery was a maximum at 70–80 ◦ C. In applications where denaturation is undesirable, selection of a gentler chromatographic mode such as ion exchange or hydrophobic interaction is recommended. 13.4 SEPARATION OF PEPTIDES AND PROTEINS 595 13.4.1.6 Capillary Columns and Nanospray Ionization Sources RPC separation of protein digests coupled on-line to nanospray MS is the preferred method for protein identification and structural characterization in proteomics studies. In most cases sample amounts are limited and very high sensitivity is required. Capillary RPC columns are used for these applications, both to achieve decreased peak volumes and to improve nanospray performance. Capillary columns with internal diameters of 50 to 150 μm can provide a theoretical 2000- to 10,000-fold improvement in sensitivity relative to a conventional 4.6-mm i.d. column. These columns are operated at flow rates of 100 to 500 nL/min, which greatly improves the efficiency of desolvation and ionization in the nanospray process. Capillary columns must be used with an ultra-low dead-volume solvent-delivery system that is capable of providing precise and accurate flow at these low flow rates, so as to achieve repeatable and otherwise acceptable gradients. Dedicated nanoflow HPLC pumps are commercially available which meet these performance requirements (Section 3.5.4.1). Alternatively, a conventional HPLC pumping system can be configured with a splitting device to reduce the flow rate (the system dead-volume downstream from the splitter must be reduced to a minimum, through the use of capillary tubing and fittings). Protein digests often contain salts which can contaminate the ionization source. Salts can be removed by installation of a peptide-trapping column and switching valve between the injector and the analytical column. Following injection, the trapping column is switched to waste to remove salts and contaminants, then switched to the analytical column for elution of peptides to the mass spectrometer. An alternative approach is the use of a vented column configuration [35]. In this system the trapping column is connected directly to the analytical column with a cross fitting. One arm of the cross serves as the high voltage connection for the nanospray source and another arm is connected to a valve that can be directed to waste (during injection and salt removal) or closed (during peptide elution). 13.4.1.7 RPC Method Development Suggested starting conditions for the development of an RPC method for a peptide or protein sample are listed in Table 13.3. If the separation using these starting conditions is inadequate, peptide separations can be improved using the strategy outlined below. Especially for the case of proteins, issues of peak shape, carryover, and reproducibility must be addressed before moving forward with separation opti- mization. Poor protein peak shape often indicates the presence of conformers or multiple species under the starting conditions. Improving peak shape may require conditions that drive the protein into a single conformation, for example, denaturing conditions as a result of elevated temperature or the use of mobile phase additives, such as surfactants. Such conditions will usually eliminate the problem of carry- over as well. Confirmation of injection-to-injection reproducibility, linearity, and elimination of carryover can be expedited by using short gradients for this purpose. Once adequate peak shape and reproducibility have been achieved, the pre- ferred strategy for optimizing the separation is to improve selectivity, for example, by changing temperature and gradient time—as described for small molecules in Section 9.3. Alternatively, computer simulation (Chapter 10) can be a more rapid and effective method for selecting the best separation conditions. In the example of . injector and the analytical column. Following injection, the trapping column is switched to waste to remove salts and contaminants, then switched to the analytical column for elution of peptides to. applications, both to achieve decreased peak volumes and to improve nanospray performance. Capillary columns with internal diameters of 50 to 150 μm can provide a theoretical 2000- to 10,000-fold improvement. acetonitrile. Solvent selectivity, cost, toxicity, and purity are also considerations. Peptides and proteins can be detected by the UV absorbance of the peptide bond at 205 to 220 nm. Acetonitrile