Anal. Chem. 2003, 75, 1931-1940
Regio-Specific Adsorption of Cytochrome c on Negatively Charged Surfaces Wensheng Xu,† Hong Zhou,‡ and Fred E. Regnier*
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907
Studies are reported on the identification of the chromatographic contact domain of equine cytochrome c during its interaction with negatively charged sorbents. A negatively charged resin was designed that would simultaneously adsorb the protein electrostatically and covalently bind it through amide bond formation to succinate groups coupled to the support in an ester linkage. Protein immobilization occurred through lysine residues participating in electrostatic adsorbed cytochrome c to the resin surface. After covalent bond formation in the interface between the protein and the sorbent, ester linkages coupling succinate groups to the support were hydrolyzed, and the protein was released. Lysine residues on the protein that had participated in covalent capture were labeled with succinate residues. The tagged protein was then tryptic-mapped and the peptides were examined by matrix-assisted laser desorption ionization mass spectrometry to determine the position of the amino acids that had been tagged. Comparing the tagged sites with the X-ray crystallographic structure of cytochrome c, it was concluded that a single face of the protein dominated the adsorption proces,s and the 3-D structure of the protein remained largely undisturbed during adsorption. Adsorption of macromolecules at the surface of chromatographic sorbents is a complex process.1 Modeling predicts that large numbers of functional groups can participate in adsorption2-7 and that three-dimensional structure can play a major role in determining how macromolecular analytes contact the surface, particularly in the case of proteins.8-12 Heterogeneity in the * Corresponding author. Fax: (765) 494-0359. E-mail:
[email protected]. † Current address: P&G Pharmaceuticals, Route 320, Woods Corners, Norwich, NY. ‡ Current address: MiniMed, Inc., 12744 San Fernando Rd., Sylmar, CA. (1) Valenzuela, D. P.; Myers, A. L. Adsorption Equilibrium Data Handbook; Prentice Hall: Englewood Cliffs, NJ, 1989. (2) Gerstner, J. A.; Bell, J. A.; Cramer, S. M. Biophys. Chem. 1994, 52, 97106. (3) Johnson, R. D.; Arnold, F. H. Biochim. Biophys. Acta 1995, 1247, 293297. (4) Johnson, R. D.; Wanf, Z.-G.; Arnold, F. H. J. Phys. Chem. 1996, 100, 51345139. (5) Melander, W. R.; Rassi, Z. E.; Horva´th, C. J. Chromatogr., A 1989, 469, 3-27. (6) Ståhlberg, J.; Jo ¨nsson, B.; Horva´th, C. Anal. Chem. 1991, 63, 1867-1874. (7) Ståhlberg, J.; Jo ¨nsson, B.; Horva´th, C. Anal. Chem. 1992, 64, 3118-3124. (8) Regnier, F. E. Science 1983, 222, 245-252. (9) Regnier, F. E. Chromotographia 1987, 24, 241-251. 10.1021/ac020335u CCC: $25.00 Published on Web 03/12/2003
© 2003 American Chemical Society
distribution of functional groups at the surface of proteins is an additional complication. This leads to the question of whether some portion of the analyte molecule dominates adsorption. Evidence that this might be the case comes from the chromatography of genetically engineered proteins varying at specific positions by one or a small number of amino acids. Single amino acid substitutions at specific positions in the three-dimensional structure of a protein have been found to impact chromatographic behavior, but similar substitutions at another sited have no effect.13-16 This has led to the “contact region” model of adsorption.17 The central hypothesis in this model is that a particular external region on the surface of a protein is most likely to interact with the surface of a chromatographic sorbent because it provides the most energetically favorable interaction. The model further asserts that the structural “footprint” or contact region would vary with the mode of chromatography. The powerful implication of this concept is that structural variations outside the chromatographic contact region would make minimal, if any, contribution to distribution equilibria and chromatographic behavior. The problem with the contact region model of adsorption is that it is based on inference. Support for the model comes primarily from correlating changes in chromatographic behavior with single amino acid substitutions in the structure of a protein.13-16 More direct evidence would be desirable. One line of direct evidence comes from a recent study in which it was shown that the pattern of peptides produced by proteolysis of cytochrome c was altered when the protein is adsorbed to a reversed-phase chromatography sorbent.18 Some peptide bonds were less likely to be cleaved when the protein was adsorbed to the reversed-phase sorbent than were found to occur in the solution. It was explained that portions of adsorbed cytochrome c not readily cleaved must spend a major portion of their time in the sorbent-analyte interface where they are less accessible to trypsin. Moreover, this would mean that adsorption of the protein to a reversed-phase sorbent must be regiospecific. The only question in this elegant approach is whether the structure of adsorbed cytochrome c changes during proteolysis. If it does, (10) Xu, W.; Regnier, F. E. J. Chromatogr., A 1998, 828, 357-364. (11) Chicz, R. M.; Shi, Z.; Regnier, F. E. J. Chromatogr., A 1986, 359, 121-130. (12) Chicz, R. M.; Regnier, F. E. J. Chromatogr., A 1988, 443, 193-203. (13) Fausnaugh, J. L.; Regnier, F. E. J. Chromatogr., A 1986, 359, 131-146. (14) Chicz, R. M.; Regnier, F. E. J. Chromatogr., A 1988, 443, 193-303. (15) Fausnaugh, J. L.; Thevenon, G.; Janis, L.; Regnier, F. E. J. Chromatogr., A 1988, 443, 221-228. (16) Chicz, R. M.; Regnier, F. E. Anal. Chem. 1989, 61, 2059-2066. (17) Regnier, F. E. Science 1987, 238, 319-323. (18) Aguilar, M. I.; Clayton, D. J.; Holt, P.; Kronina, V.; Boysen, R. I.; Purcell, A. W.; Hearn, M. T. W. Anal. Chem. 1998, 70, 5010-5018.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003 1931
basic amino acid cleavage sites could be either exposed or buried during structural rearrangement. This would give a distorted picture of the structure of adsorbed proteins. An additional question is the extent to which the contact region will vary between chromatographic modes. Will it be the same for other modes of chromatography as with reversed phase? The work reported here was directed at determining the chromatographic contact region of cytochrome c on a cation exchange chromatography sorbent. This question was addressed by forming covalent bonds between functional groups on the surface of the protein and the sorbent while the protein was adsorbed. The study was executed with a “venus flytrap” type of sorbent specially synthesized for the purpose of identifying the chromatographic contact region. Trapping groups on the sorbent surface were designed to attract cationic proteins electrostatically and then covalently capture them through amide bond formation. Primary amine groups in the contact region would have the highest probability of capture. After covalent capture, cytochrome c was released from the surface with succinate groups involved in covalent linkage to the sorbent surface still appended to lysine residues involved in the capture process. Protein thus derivatized was sequenced to locate sites of succinate attachment and lysine residues that participated in adsorption. EXPERIMENTAL SECTION Chemicals and Reagents. Cytochrome c (from equine heart), trypsin (TPCK treated), 1,3-dicyclohexyl carbodiimide (DCC), ethylene glycol, TRIZMA hydrochloride (Tris-HCl) and TRIZMA base (Tris base, 99.9%) were purchased from Sigma Chemical Company (St. Louis, MO). POROS hydroxylated polystyrene divinylbenzene (PS-DVB-OH) was a generous gift from PerSeptive Biosystems (Framingham, MA). Sulfo-N-hydroxysuccinimide (sulfo NHS) and trifluoroacetic acid (TFA) were purchased from Pierce Chemical Company (Rockford, IL). Pyridine, dimethylaminopyridine (DMAP), dimethysulfoxide (DMSO), dioxane and ethanol amine, R-cyano-4-hydroxycinnamic acid (CHCA), and sinapinic acid (SA) were purchased from Aldrich Chemical Co. (Milwaukee, WI). Urea, sodium bicarbonate, and ammonium bicarbonate were purchased from J. T. Baker (Phillipsburg, NJ). HPLC grade acetonitrile (ACN) was purchased from E. M. Science (Gibbsburg, NJ). Tetrahydrofuran (THF) was obtained from Fisher Scientific (Fair Lawn, NJ). Apparatus. All chromatographic separations were performed on a BioCAD liquid chromatograph (PerSeptive Biosystems, Framingham, MA). Absorbance was measured at 214 nm. Chromatographic separations were performed at ambient temperature. Mass measurements were performed on a Voyager mass spectrometer (PerSeptive Biosystems, Framingham, MA) in the positive ion, linear mode for proteins and the reflector mode for peptides. Chromatographic Evaluations. Tryptic peptide separations were performed on a 250 × 4.6 mm i.d. column packed with C-18 reversed-phase, 5-µm particle size sorbent of 300-Å pore size obtained from PerSeptive Biosystems (Framingham, MA). Bulk solvents were filtered through a 0.45-µm nylon filter and degassed with helium. A linear gradient was performed from 100% solvent A (0.1% TFA in 1% ACN and 99% water) to 100% solvent B (0.1% TFA in 95% ACN and 5% water) for 60 min at a flow rate of 1 1932
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
mL/min. A 100-µL portion of sample was loaded for each separation. The buffer solution was prepared in double-distilled water that had been passed through a 0.45-µm nylon filter. Mass Spectrometry Analysis. Matrixes for the analysis of peptides and proteins were R-cyano-4-hydroxycinnamic acid (CHCA) and sinapinic acid (SA), respectively. Two microliters of 50% acetonitrile in 0.3% aqueousTFA (matrix A diluent) was added to peptide samples to produce a final peptide concentration of ∼1 µM. Two microliters of 30% acetonitrile in 0.3% TFA (matrix B diluent) was added to the protein samples to produce a final protein concentration of ∼1 µM. Samples (0.5 µL) were spotted on a stainless steel MALDI plate for analysis. Reaction of POROS PS-DVB-OH with Succinic Anhydride. A three-necked round-bottom flask fitted with a mechanical stirrer, a nitrogen inlet, and a gas outlet was charged with 1 g of POROS PS-DVB-OH, 0.1 g of DMAP, 0.5 g of succinic anhydride, and 30 mL of anhydrous pyridine. The suspension was stirred for 20 h, and the resulting solution of light brown coloration was filtered through a suction filter to remove solvent from the particles. The solids were washed with acetonitrile and 0.1 M HCl. Finally, this succinylated resin was washed with deionized water and dried under vacuum for 10 h before activation. Determining the Surface Density of Succinate Residues. A 50-mL round-bottom flask containing 10 mL of dry pyridine was charged with 0.1 g of succinylated PS-DVB-OH resin, 12 mg of trichloroethanol, and 17 mg of dicyclohexyl carbodiimide. The mixture was agitated gently at room temperature for 12 h, after which the esterified resin was isolated by filtration and washed with THF. The 2,2,2-trichlorethanoyl succinate PS-DVB-OH resin was then dried under vaccum for 10 h and sent for elemental analysis, where it was found to have a chlorine content of 160 µmol/g. Activation of Succinylated Resin with Sulfo-N-hydroxysuccinimide (Sulfo-NHS). Seventy mg of sulfo-NHS was dissolved in 5 mL of water and added to 1 g of the succinylated resin in a 50-mL Erlenmeyer flask. The reaction vessel was agitated at room temperature for 30 min. Subsequently, 65 mg of DCC in 5 mL of dioxane was added, and the flask was agitated for another 2 h at room temperature. The activated sorbent thus produced was then isolated by filtration and washed thoroughly with dioxane and water. The sorbent was then dried and used immediately. Covalent Coupling of Cytochrome c to Sulfo-NHS-Activated Surfaces. Ten milliliters of 6 mg/mL cytochrome c solution in 0.1 M NaHCO3 (pH 7.5) was added to 1 g of sulfo-NHS-activated sorbent in a 50-mL test tube and gently agitated at room temperature for 16 h. After filtration through Celite, unreacted sulfo-NHS ligands were deactivated by adding 5 mL of a 0.1 M solution of ethanolamine in 0.1 M NaHCO3 (pH 7.5). After agitation at room temperature for another 2 h, the sorbent was isolated by filtration and subjected to thorough washings sequentially with water, 1 M NaCl, 10% ethylene glycol, and water again. Release of Bound Protein. One gram of sorbent with immobilized protein and 5 mL of 0.05 M Tris buffer (pH 8) were added to a test tube, and the pH was adjusted to 12 with 1 M NaOH. The tube was strongly agitated at 45 °C for 10 h. Adjustments were made to maintain the pH, if necessary. At the end of 10 h, the solution was neutralized with 1 M HCl and then filtered through Celite. The filtrate contained the red-colored
Figure 1. Procedure for synthesis of the negatively charged surface and isolation of tagged protein.
succinylated cytochrome c released from the sorbent. After refiltration through a 0.2-µm filter, the succinylated cytochrome c was lyophilized before analysis. Free Solution Tryptic Digest. Two milligrams of either cytochrome c or modified cytochrome c was dissolved in 0.4 mL of 6 M urea and 0.4 mL of 50 mM Tris (pH 8.0) with 1 min of vortex stirring. The final solution contained 1 mg/mL cytochrome c or modified cytochrome c in 2 mL of 1.2 M urea and 50 mM Tris (pH 8.0). Forty microliters of trypsin (TPCK treated) at a concentration of 0.5 mg/mL was added to 2 mL of 1 mg/mL cytochrome c in a ratio of 1 to 100 by weight and incubated at 37 °C. After 8 h, a second aliquot of trypsin was added. Following a total incubation time of 24 h, samples were frozen with liquid nitrogen and stored frozen until analyzed. RESULTS AND DISCUSSION Design of a “Venus Fly Trap” Sorbent. The term “venus fly trap” is used to refer to sorbents that electrostatically attract and then covalently capture proteins. [“Venus fly trap” is derived from a flower of this common name that attracts insects and then captures them.] This study was based on the concept that the contact region of a protein adsorbed to an ion exchange column could be identified by (1) electrostatically attracting the protein to the charged surface of the sorbent, (2) covalently binding it to the sorbent while it was adsorbed through functional groups in the interface, and (3) subsequently identifying functional groups involved in these covalent linkages to the surface through conventional protein structure analysis. A cation exchange sorbent capable of covalently binding primary amines electrostatically attracted to the surface was prepared by first grafting succinic acid to the surface of a hydroxylated support through an ester linkage (step I of Figure 1). The free carboxyl group of the resulting succinate half ester was then activated with sulfo-N-hydroxysuccinimide to form a strong cation exchange (SCX) sorbent (step II of Figure 1). Equine cytochrome c with a net positive charge (pI ) 10) was electrostatically adsorbed to the negatively charged sorbent. After adsorption, lysine residues in the sorbent-protein interface nucleophilically displaced sulfo-N-hydroxysuccinimide from the sorbent and were
covalently coupled to the resin surface (step III). Lysine amino groups spending the greatest time electrostatically adsorbed at the sorbent surface would be the most likely to form covalent bonds to the support. Cytochrome c thus bound to the sorbent was subsequently released by hydrolysis of the ester linkage between the support and the succinate coupling groups. Hydrolysis was achieved under weakly basic conditions. The released protein carried succinate tags at all of the amino groups participating in covalent attachment to the surface. Amino groups involved in binding the protein to the support were located in two ways. One was to determine which amino acids in the released protein carried succinate tags by tryptic mapping and mass spectral analysis. Comparing the tryptic map of the derivatized protein to that of native cytochrome c revealed those peptides that had been derivatized and the number of succinate tags on the peptide. The number of succinate moieties attached to a modified tryptic peptide was determined by matrix-assisted laser desorption ionization mass spectrometry (MALDI-MS). The total number of sites modified in the protein was also determined by MALDI-MS, using the intact protein before proteolysis. A second strategy for identification of binding sites was to trypsinize the immobilized protein directly without releasing it from the sorbent. Peptides found in tryptic digests of the unbound protein that did not appear in the tryptic map of the bound protein were assumed to be involved in covalent binding to the sorbent surface. This approach was less effective, and the data will not be presented. Characterization of the “Venus Flytrap”. Charge density is an important issue in any ion exchange sorbent. Because the sulfoN-hydroxysuccinimide (sulfo-NHS) ion exchanger was quantitatively attached to succinate groups at the support surface, charge density would be equal to the density of succinate groups coupled to the support. The density of succinate residues on the support surface was determined in two steps. In the first, succinate carboxyl groups were esterified with trichloroethanol in a dicyclohexyl carbodiimide catalyzed coupling. After this esterification step, the chlorine content of the derivatized support was determined by elemental analysis and found to be 160 µmol/g. Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1933
Figure 2. Heme packing diagram of horse cytochrome c with arrows showing theoreical tryptic cleavage. Lysine residues are shown in gray ovals. The symbols T-1 through T-22 identify the tryptic peptides derived from cytochrome c.
The uncoated poly(styrene-divinylbenzene) support used in these studies had a surface area of 70 m2/g according to the manufacturer. Using the value of 160 µmol/70 m2, it can be calculated that each succinate residue occupied 0.73 nm2. This translates into an average distance between succinate residues of roughly 8.5 Å. It is important to note that 8.5 Å is an average; some groups will be closer because of the random distance between hydroxyl groups in the polymer coating to which succinate residues were attached. Total Number of Binding Sites. Equine cytochrome c is composed of 104 amino acids (Figure 2), of which there are 19 lysine residues, 2 arginine residues, and 12 acidic residues from aspartic and glutamic acid. The primary structure is a single, continuous polypeptide chain with no disulfide bridges. Direct analysis of the released protein by MALDI-MS showed that an average of 11 out of the 19 lysine residues in the protein were succinylated (Figure 3). This conclusion is based on subtracting the observed relative mass of the succinylated protein (Mr ) 13463.7) from that observed with the equine cytochrome c starting material (Mr ) 12361.7) and dividing by the relative mass of the succinate derivatizing agent (100 amu). It should be noted that the mass peak for the succinylated protein in Figure 3 is broader than that of native equine cytochrome c. This is interpreted to mean that the derivatized protein is a mixture of unresolved succinylated species of which the 11-residue modification is the most abundant. Attempts to resolve these succinylated cytochrome 1934 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
c species by anion exchange chromatography also failed. The derivatized protein mixture eluted in a broad, unresolved peak during gradient elution (data not shown). One of the questions raised by these data is why a mixture was formed instead of a single species. The adsorption and covalent bonding process is envisioned as occurring in the following way. Prior to and immediately after electrostatic adsorption the analyte is able to both diffuse and precess at the sorbent surface. Covalent binding at the first lysine residue is seen as a critical event. Once covalent coupling to the surface occurs at any point in the protein, molecular motion is severely restricted. Adjacent lysine residues would then bind covalently to juxtapositioned sulfo-N-hydroxysuccinimide groups on the sorbent surface. Covalent binding would continue to spread from this initiation site until intramolecular structural constraints would limit the further approach of primary amino groups in protein to the surface. If this picture of the process is true, why was there so much heterogeneity in attachment and derivatization, as seen above and in cases to be described below? One reason could be that the sorbent surface also introduces steric constraints into the binding process, in addition to being sterically heterogeneous. The average distance between sulfo-N-hydroxysuccinimide groups on the surface is 8.5 Å, but this distance probably ranges to greater than 10 Å in some cases. When stationary phase ligand density at a site is heterogeneous and greater than the distance between lysine residues in the protein, some lysine residues in the protein-
Figure 3. Mass spectra of (a) native cytochrome c and (b) succinylated cytochrome c, showing the maximum number of succinylation is 11.02.
sorbent interface will not come in contact with a sulfo-NHS residue and be bound. Another reason for derivatization heterogeneity could be that initial binding occurs at the outer limits of the chromatographic contact region. If covalent coupling occurs first on the outside of a charged domain, molecular motion might be sufficiently restricted to limit adjacent lysine residues from approaching the surface. The second question raised is whether the results in Figure 3 are reasonable for a protein with native structure. This was examined using a displacement model of ion exchange chromatography that predicts the number of sites in a protein involved in electrostatic adsorption.19 According to the model, the slope (Z) of a log k′ versus log 1/[D] plot is the number of charged sites on the protein interacting with the sorbent when D is salt concentration in the mobile phase. It is expected that the Z value of a sorbent and the maximum number of succinate groups found in the protein should be related, but not necessarily identical. Chromatography is a dynamic process that occurs rapidly, whereas covalent bond formation is irreversible.10,20 Changes in protein conformation could occur during the lengthy process of covalent bond formation. Although the sulfo-N-hydroxysuccinim(19) Kopaciewicz, W.; Rounds, M. A.; Fausnaugh, J.; Regnier, F. E. J. Chromatogr., A 1983, 266, 3-21. (20) Rarnayake, C. K.; Regnier, F. E. J. Chromatogr., A 1996, 743, 25-32.
ide sorbent is a cation exchanger, it cannot actually be used for chromatography. The theoretical number of binding sites for equine cytochrome c was determined with weak cation exchange sorbent10,20 and found to be 8 ( 1. Although Z values for a particular protein can vary between sorbents, the observed Z value and average number of succinate residues found in the derivatized protein are in the same range. Analysis of Tryptic Peptides. The tryptic peptide map of succinate tagged cytochrome c is very different from that of native equine cytochrome c (Figure 4 A,B). This is attributed to two facts. One is that lysine residues derivatized with succinate cannot serve as trypsin cleavage sites. This means that two or more “tryptic peptides” will remain coupled at sites of succinylation, generally appearing at longer retention time in reversed-phase chromatography (Figure 4B) and at higher mass. The second reason for the complexity of the map is that cytochrome c is randomly succinylated at multiple sites, as the data below will show. Many different succinylation patterns are possible. Fractions collected from reversed-phase chromatography (Figure 4B) were subjected to MALDI-MS (Table 1). Succinylated lysine residues are listed in Table 1, for example, 2ST1-3, indicating two succinylated lysine residues, whereas underivatized lysine residues are indicated by for example, T1-3. The sequences assigned to the observed masses of peptides are based on (i) the Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1935
Figure 4. Reversed-phase chromatogram of tryptic digested peptide of (a) cytochrome c and (b) tagged cytochrome c.
known specificity of trypsin for underivatized lysine and arginine residues, (ii) the fact that trypsin is an endopeptidase that occasionally leaves two basic amino acids at the C-terminus, (iii) the addition of 100 amu to the mass of lysine residues that were potentially succinylated, and (iv) the prospect that proteolysis was not complete. The position of succinylated amino acids in the protein sequence and the number of succinate residues in the 1936 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
derivatized peptides are seen in Table 2. It should be noted that the same derivatization position could be found in 4-6 different peptides. This is further proof of the randomness of derivatization. Numbers of succinylation sites on peptides ranged from 1 to 7. The fact that the average number of succinate residues found in the protein released on ester hydrolysis was 11 means that multiple succinylated peptides were generated upon proteolysis.
Table 1. Free Solution Tryptic Digested Fragments of Succinylated Cytochrome c peptidea T1-3 2ST1-3 T4 1ST1-4 1ST3-4 T5 2ST3-5 1ST4-5 2ST1-6 T8 1ST1-8 T2-8 2ST6-8 T7-8 1ST7-8 7ST1-9 T3-9 T5-9 2ST6-9 1ST7-9 T8-9 T10 T9-10 1ST9-10 T9-11 1ST9-11 T10-11 1ST10-11 T12 T10-12 T13 T11-13 T13-14 T15 T14-15 T16 2ST14-16 3ST13-17 4ST13-17 3ST14-17 1ST16-17 2ST16-18 T14-19 2ST14-19 5ST14-19 T15-19 1ST16-19 2ST16-19 3ST16-19 1ST18-20 T19-20 T19-21 1ST20-21 T17-22 3ST17-22 1ST18-22 2ST18-22 3ST18-22
sequence
Mrb/∆amuc
GDVEKGKK GDVEKGKK IFVQK GDVEKGKKIFVQK KIFVQK CdAQCdHTVEK KIFVQKCdAQCdHTVEK IFVQKCdAQCdHTVEK GDVEKGKKIFVQKCdAQCdHTVEKGGK TGPNLHGLFGR GDVEKGKKIFVQKCdAQCdHTVEKGGKHKTGPNLHGLFGR GKKIFVQKCdAQCdHTVEKGGKHKTGPNLHGLFGR GGKHKTGPNLHGLFGR HKTGPNLHGLFGR HKTGPNLHGLFGR GDVEKGKKIFVQKCdAQCdHTVEKGGKHKTGPNLHGLFGRK KIFVQKCdAQCdHTVEKGGKHKTGPNLHGLFGRK CdAQCdHTVEKGGKHKTGPNLHGLFGRK GGKHKTGPNLHGLFGRK HKTGPNLHGLFGRK TGPNLHGLFGRK TGQAPGFTYTDANK KTGQAPGFTYTDANK KTGQAPGFTYTDANK KTGQAPGFTYTDANKNK KTGQAPGFTYTDANKNK TGQAPGFTYTDANKNK TGQAPGFTYTDANKNK GITWK TGQAPGFTYTDANKNKGITWK EETLLMEYLENPK NKGITWKEETLLMEYLENPK EETLLMEYLENPKK YIPGTK KYIPGTK MIFAGIK KYIPGTKMIFAGIK EETLLMEYLENPKKYIPGTKMIFAGIKK EETLLMEYLENPKKYIPGTKMIFAGIKK KYIPGTKMIFAGIKK MIFAGIKK MIFAGIKKK KYIPGTKMIFAGIKKKTER KYIPGTKMIFAGIKKKTER KYIPGTKMIFAGIKKKTER YIPGTKMIFAGIKKKTER MIFAGIKKKTER MIFAGIKKKTER MIFAGIKKKTER KTEREDLIAYLK TEREDLIAYLK TEREDLIAYLKK EDLIAYLKK KKTEREDLIAYLKKATNE KKTEREDLIAYLKKATNE KTEREDLIAYLKKATNE KTEREDLIAYLKKATNE KTEREDLIAYLKKATNE
861.0/1.0 1059.2/-0.8 634.5/0.7 1576.6/0.8 861.3/-0.7 1633.0/4.9 2577.1/5.0 2349.0/5.1 3532.5/4.3 1168.5/0.1 4850.5/6.6 4721.8/6.5 1875.7/-0.3 1433.9/0.2 1533.6/-0.1 5469.4/-2.7 4157.9/-0.4 3418.9/4.6 2004.4/0.2 1662.1/0.2 1296.6/0.0 1477.0/0.4 1599.1/0.3 1699.3/0.4 1841.4/0.3 1941.5/0.4 1714.2/1.3 1813.3/0.4 604.4/0.7 2298.9/0.3 1495.8/-0.1 2326.5/2.6 1624.0/-0.1 678.5/0.7 805.6/-0.4 779.5/0.5 1766.9/-0.1 3475.4/2.3 3576.9/3.8 1995.0/-0.2 1007.7/0.5 1235.6/0.2 2209.0/-0.9 2407.5/-2.8 2711.6/1.7 2081.5/-0.2 1521.7/-0.2 1621.7/-0.2 1721.7/-0.2 1704.2/-2.9 1478.6/-0.3 1606.7/-0.4 1322.0/1.4 2276.6/-2.3 2577.4/-1.5 2249.4/-1.3 2349.0/-1.7 2451.9/1.2
position 1-8 9-13 1-13 8-13 14-22 8-22 9-22 1-25 28-38 1-38 6-38 23-38 26-38 1-39 8-39 14-39 23-39 26-69 28-39 40-53 39-53 39-55 39-55 40-55 56-60 40-60 61-72 54-72 61-73 74-79 73-79 80-86 73-86 61-87 61-87 80-87 80-87 73-91 74-91 80-91 88-100 89-99 89-100 92-100 87-104 88-104
a The peptide fragment and succinylated peptide fragment, for example, T1-3, is a native peptide from fragment 1 to 3; 2ST1-3 contains 2 succinylated lysine from fragment 1 to 3; succinylated lysine could be any two of lysines at amino acid position 5, 7 and 8. b Measured molecular mass (Da). c The difference between measured molecular mass and theoretical mass. d The heme prosthetic group is covalently attached to cysteine-14 and cysteine-17.
Another way to examine the tryptic peptide data is to plot succinylation frequency relative to sequence position in the primary structure of the protein (Figure 5). The sequence of the peptide within which the succinate was found is ignored. In effect, this measures the propensity of a given position in the protein to be involved in binding. It is clear that two regions in the primary
structure ranging from residues 5 through 27 and 73 through 88 have the highest propensity to be involved in binding. The question of where these regions are in the three-dimensional structure of the protein will be explored below. The limited activity of some regions is also evident. The large differences in reactivity of lysine residues in close proximity, such as residues 22 and 25 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1937
Table 2. List of Succinic Acid Numbers on the Lysines lysine position on cytochrome c succinylated peptide T1-3 T1-4 T1-6 T1-8 T2-8 T1-9 T3-4 T3-5 T4-5 T6-8 T6-9 T7-9 T7-8 T9-11 T9-10 T10-11 T13-17 T14-16 T14-17 T14-19 T16-17 T16-18 T16-19 T17-22 T18-22 T18-20 T20-21
5
7
X X X X
x
X
8
13
22
25
27
x x x x
x x
x x
x x
x
39
53
55
60
72
73
79
86
x
x x x x
x x x x
x
87
88
x
x
99
100
x
x
x x x
x x x x
x x x x x x
Figure 5. Succinic acid appearance on the lysines.
or 72 and 73, are surprising. Another striking fact is that positions 55 and 60 were never derivatized. The prospect that the amino terminus was succinylated and the tryptic peptide was not retained on the reversed phase column can also not be eliminated. Data Interpretation in Terms of 3-D Structure. The X-ray crystallographic structure of equine cytochrome c is well established.21-24 Two lysine-rich domains are found on a single (21) Dickerson, R. E.; Takano, T.; Eisenberg, D.; Kallai, O. B.; Samson, L.; Cooper, A.; Margoliash, E. J. Biochem. 1971, 246, 1511-1535. (22) Matsuura, Y.; Takano, T.; Dickerson, R. E. J. Mol. Biol. 1982, 156, 389409. (23) Trewhella, J.; Carlson, V. A.; Curtis, E. H.; Heidorn, D. B. Biochemistry 1988, 27, 1121-1125. (24) Voet, D.; Voet, J. Biochemistry, 2nd ed.; John Wiley & Sons: New York; pp 579-1995.
1938 Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
x x x x x x
x x x
x x x x
x
face of the protein. Domain I consists of residues 8, 13, 25, and 27, whereas domain II is a long, curving loop containing residues 73, 79, 86, 87, and 88. Together, these 9 residues encircle the heme moiety in the protein. It is clear to see that this cationic face of the protein dominates adsorption in cation exchange chromatography. But Figure 5 shows that adjacent to domain I, residues 5 and 7 are also strongly involved, while residue 22 makes a smaller contribution. The X-ray crystallographic structure of equine cytochrome c shows that residues 5 and 7 are in an R helix projecting away at a right angle from the face of the protein containing residues 8, 13, 25, and 27 (Figure 6). How, then, are they involved in bond formation? It would be necessary that this amino-terminal portion of the protein undergo a conformational change, bringing residues 5 and 7 into the sorbent-protein interface. Apparently, this occurs, and residues 5, 7, 8, 13, 25, and 27 are brought into roughly the same plane by this conformational alteration. X-ray data also suggest an explanation for the diminished participation of residue 22. This lysine projects away from the 8, 13, 25, 27 face and could be slightly further from the sorbent surface than adjacent residues 25 and 27. Residues 39 and 53 are at the periphery of domain I. Minimal participation of these residues can also be explained in terms of steric limitations. If they occupy the same position in the adsorbed protein as shown in Figure 6, they are limited in their approach to the sorbent surface. Steric effects imposed by the structure of cytochrome c can also be used to explain the participation of amino acid residues in domain II. Residues 99 and 100 are at the periphery of the domain and lie a short distance behind the plain of the interface. Even residues 73 and 79 lie slightly behind the plain. This might
Figure 6. (a) 3-dimensional structure of cytochrome c. Eleven positions labeled with numbers are the most probable binding sites. (b) Cytochrome c sequence.
explain the small reduction in their propensity to bond to the sorbent. Lysine residues 55 and 60 are buried deep within the 3-D structure, according to X-ray crystallography, and therefore are not readily accessible for reaction, as indicated in Figure 5. They are obviously not in the contact region involved in cation exchange chromatography. Studies of cytochrome c adsorption to a reversed phase sorbent16 showed that the N-terminal lysine 22, as well as C-terminal residues 60, 72, 73, 79, 87, 99, and 100 and arginine 91, were inaccessible to tryptic digestion when the protein was adsorbed. It was concluded that the sequence region encompassing glycine 56 to glutamate 104 comprises the chromatographic contact region in reversed-phase chromatography. This is clearly a very different region from residues shown here to be involved in cation exchange chromatography. Correlation of X-ray crystallographic structure data with the behavior of proteins adsorbed on chromatographic surfaces is valid only if the structure is the same in the two systems. The structure of proteins adsorbed on ion exchange and reversed-
phase materials has also been examined by Raman spectroscopy25 and nuclear magnetic resonance spectroscopy with isotopeexchange techniques.26 Neither ribonuclease A, a rigid protein, nor lactalbumin, a flexible protein, exhibited any significant secondary structural change on adsorption to an agarose-based cation-exchange support. These reports suggest that the structure of cytochrome c, too, will not be substantially perturbed upon adsorption to the surface of a cation exchanger. Reversed-phase supports, however, can be different. Lysozyme showed a significant perturbation in secondary structure in the adsorbed state, as compared to it’s structure in solution.26 But once adsorption had occurred, the perturbed structure was relatively insensitive to mobile phase composition. CONCLUSIONS It may be concluded that adsorption of cytochrome c in cation exchange chromatography is due to adsorption at multiple (25) Sanel, S. U.; Cramer, S. M.; Przybycien, T. M.; Isermann, H. P. J. Chromatogr., A 1999, 849, 149-159. (26) McNay, J. L.; Fernandez, E. J. J. Chromatogr., A 1999, 849, 135-148.
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
1939
residues and is dominated by two structural domains on a single face of the protein involving predominantly cationic residues 5, 7, 8, 13, 25, 27 in domain I and 73, 79, 86, 87, and 88 in domain II. It is further concluded that during the course of adsorption, a small conformational change occurs involving residues 5 and 7 at the amino terminus to bring them into the sorbent-protein interface. The results reported here lend credibility to the hypothesis that only a portion of the external surface of a protein can dominate its chromatographic behavior. Furthermore, it appears that the contact region might not be the same for different modes of
1940
Analytical Chemistry, Vol. 75, No. 8, April 15, 2003
chromatography. The significance of these findings is that they explain why it is that single amino acid variants may be resolved by one type of chromatography, but not another. ACKNOWLEDGMENT The authors gratefully acknowledge financial support from the National Institute of Health (GM 59996). Received for review May 20, 2002. Accepted January 31, 2003. AC020335U
